National Fire Incident Reporting System Research Articles

The burden of natural gas leaks on public sector emergency response in the United States

Fire departments play a critical role in responding to potential hazards generated by natural gas and propane leaks, yet the costs of providing these services is unknown. Here, we analyzed 15 years of gas leak-coded incidents from the National Fire Incident Reporting System (NFIRS) and estimated associated costs using fire department operating budgets. From 2003 to 2018, a total of 2.4 million gas leak-coded incidents required fire department responses, accounting for ∼0.8% of all nationwide incidents. Reported gas leak incidents have steadily increased nationwide, nearly quadrupling from 2003 to 2018. In 2018, New York City (NYC) experienced 22,090 gas leak incidents—more than the next 25 cities combined—conservatively costing NYC $70 million annually. With a median national response cost of $2609 (S.D. $1734), uncombusted gas leaks conservatively cost U.S. fire departments $564 million in 2018—at least 10x greater than for gas leak-caused fires. City-level natural gas systems entail an underappreciated and increasing public cost burden on emergency response services. Findings herein raise the salience of gas leak-associated emergency response costs and support improved gas leak response coordination between emergency services and gas utility providers. Decision-makers are encouraged to incorporate these costs into cost-benefit analyses and consider the intended and unintended effects on emergency response systems in managing energy transitions.

A novel historical data-based method for predicting firefighters demand in urban fires

Urban fires are a prevalent type of fire that requires prompt and effective responses. Determining the appropriate number of firefighters at the alarm time is crucial, but it mostly relies on the experience of decision-makers. Due to the fuzziness and limitations of human knowledge, there is often a deviation between the number of firefighters dispatched relying on personal subjective experience and the realistic demand. This paper proposed a historical data-based method for predicting the demand number of firefighters in urban fire. Initially, the National Fire Incident Reporting System (NFIRS) data and Global Historical Climatology Network Daily (GHCN-D) weather data were combined to create a fused dataset. The dataset was subjected to anomaly detection and feature selection, then the processed data was used to predict the number of firefighters using artificial neural network (ANN). In this process, the Genetic Algorithm (GA) was applied to optimize ANN structure. Finally, comparative experiments were conducted to evaluate the performance of the proposed method, which demonstrated superior accuracy in comparison to common regression models and traditional ANN. This advancement brings the number of dispatched firefighters closer to the actual demand and contributes to ensuring an adequate allocation of firefighters for urban fires. Consequently, decision-makers can make more informed decisions regarding the number of firefighters to dispatch, leading to more effective responses to urban fires.

Examining Exposure Fires from the United States National Fire Incident Reporting System between 2002 and 2020

Fires resulting from antecedent fires, known as exposure fires, can manifest across diverse environments, including suburban, urban, and rural areas. Notably, exposure fires represented by structure-destroying fires within the wildland–urban interface (WUI) can extend into non-WUI suburban and urban regions, presenting significant challenges. Leveraging data from the United States National Fire Incident Reporting System (NFIRS) spanning 2002 to 2020, this study investigates 131,739 exposure fire incidents impacting 348,089 features (incidents). We analyze reported economic costs, affected feature types, and property utilization patterns for these exposure fires. We also compare these exposure fires to information documented in other databases. Finally, we examine structure separation distance at residential dwellings and describe ignition pathways for selected fires. Reported property losses for some fire incidents amounted to USD 5,647,121,172, with content losses totaling USD 1,777,345,793. Prominent fire incident categories include buildings, vehicles, and natural vegetation fires, predominantly occurring in residential, outdoor, and storage areas. While the NFIRS lacked information on most major structure-destroying WUI fires, highlighting this analysis’s lack of statistical representation, it did provide insights into less extensive exposure fires, both WUI and non-WUI, unrecorded elsewhere. Our study reveals significant distinctions in the distribution of separation distances between damaged-to-damaged structures (average separation of 6.5 m) and damaged-to-not-damaged structures (average separation of 18.1 m). Notably, 84% of the incidents in exposure fires involved fire suppression defensive actions. These defensive actions contributed to the differences in structure separation distance distributions, highlighting the often-neglected role of these measures in assessing structure responses during WUI fires. We examined ignition pathways at select exposure fires, highlighting some common features involved in fire spread and challenges in documenting these pathways. Finally, we propose a set of idealized attributes for documenting exposure fires, accentuating the inherent difficulties in collecting such data across expansive geographical areas, particularly when striving for statistical representation. Our findings yield valuable insights into the multifaceted nature of exposure fires, informing future research and database development to aid in mitigating their impact on vulnerable communities.

Interpretable boosting tree ensemble method for multisource building fire loss prediction

Building fires may cause enormous property loss. Disaster relief organizations and post-disaster recovery efforts benefit from the accurate and interpretable prediction of property loss. To solve this problem, we propose a novel interpretable boosting tree ensemble method (IBTEM), which is of practical significance for providing decision support for dispatching aid and mobilizing recovery resources. First, we fuse multisource datasets including National Fire Incident Reporting System (NFIRS) dataset and National Oceanic and Atmospheric Administration (NOAA) dataset from 2012 to 2016. Second, we construct four variable scenario subsets to select related variables for building fire loss. Third, we adopt Winsorization, logarithmic transformation, recursive feature elimination and weighted voting strategies to create an ensemble of boosting trees. Fourth, we conduct interpretable Shapley additive explanations to analyze the model internal mechanism. The proposed IBTEM is compared with other popular machine learning methods and the experimental results show the IBTEM achieves outstanding superiority. Property value, fire spread and number of stories with minor damage are verified the most effective variables for loss prediction. In conclusion, the IBTEM realizes accurate and interpretable loss prediction of building fires, and assists relevant departments in making disaster relief decisions in a timelier manner.

Analyzing the national fire incident reporting system to identify carbon monoxide incidents in the U.S. lodging industry

Carbon Monoxide (CO) poisonings in the U.S. lodging industry have become a regular occurrence, however there is no current mandatory national reporting, tracking, or surveillance mechanism for CO incidents in the U.S. lodging industry. As such, the problem is largely invisible. The objective of this study was to utilize the National Fire Incident Reporting System (NFIRS) data to better understand the public health risk from Carbon Monoxide (CO) incidents in the U.S. lodging industry. The NFIRS datasets for years 1999 through 2018 were assessed to identify CO incidents occurring at U.S. hotels and motels. The results of the analysis identified 3405 incidents. Incidents were strongly correlated with increased fire department participation in NFIRS (R = 0.82). The number and frequency of CO incidents in the U.S. lodging industry are underreported. Previous efforts relying on news media identified only 10% of the incidents reported in the NFIRS data. This indicates a greater public health risk associated with CO exposure in the U.S. lodging industry than previously realized.

Combining spatial and sociodemographic regression techniques to predict residential fire counts at the census tract level

This work examines different spatial and sociodemographic models for predicting residential fire counts at the census tract level for 118 U.S. fire departments across 25 states. The models give five-year forecasts of residential fire counts for 3392 census tracts which contain over 13 million residents in total. All models described in this paper train on fire incident data from the National Fire Incident Reporting System (NFIRS) over the interval 2006–2011 (inclusive) and are evaluated based on their ability to predict the fire counts that occurred over the interval 2012–2016. Two strictly spatial models are considered- a simple “count” model that serves as a baseline for all other models described in the paper and a model that utilizes kernel density estimation (KDE) with statistically optimized bandwidths. Using data from the American Community Survey (ACS), an examination of the effects of demographic and housing factors on the fire risk is presented. The data suggest that the fire risk per person is generally higher in census tracts with attributes corresponding to socioeconomic disadvantage such as low median incomes and small fractions of residents with college degrees. These trends inform the design of a Bayesian hierarchical Poisson regression model, which is shown to make predictions with a 9% lower root mean squared error (RMSE) relative to the base model. A spatial kernel regression is then conducted on the residuals of this regression, which results in a 15% RMSE improvement relative to the base model. These results are compared to a conditional autoregressive (CAR) model, which incorporates spatial information directly into the hierarchical Poisson regression. Although the RMSE is higher for the CAR model's point estimate forecasts (7% lower than the base model), it allows for the generation of probabilistic forecasts and gives spatially-informed statistical estimates of the effects of the sociodemographic variables. This work highlights the utility of geocoded fire incident and demographic data as well as machine learning techniques that can utilize these datasets to make improved predictions.

A hybrid ensemble learning method for the identification of gang-related arson cases

Arson is one of the most common crimes, and it has the characteristics of low cost and great harm. In addition to causing casualties and property damage, arson can often have huge social impacts and cause psychological panic in the public. Since arson is more harmful when conducted by a gang, how to effectively identify gang crimes in arson cases has become an important issue. In this paper, we propose a hybrid method that combines ensemble learning and intelligent optimization algorithms to solve this problem. First, we develop the recursive feature elimination (RFE)-based feature selection method to remove redundant features. Second, for the data imbalance problem, we determine the optimal processing algorithm from 18 candidate algorithms. Third, after trying a combination of multiple base classifiers, we obtain the optimal base classifier combination. Fourth, when integrating the prediction results of the base classifier, we propose a weighted ensemble strategy. Finally, we use the differential evolution (DE) algorithm to optimize the parameters of the base classifier and the weight of the combination, which further enhances the identification ability of the model. To verify the actual performance of the proposed method, we conducted experiments on the US National Fire Incident Reporting System (NFIRS) database. The results show that the proposed method is significantly superior to other popular machine learning methods, which proves that this method can provide a more reliable decision basis in the detection of arson cases.

Soft furnishing fires: They're still a problem

SummaryHome fires in which soft furnishings were first ignited or contributed to fire spread cause a disproportionate share of US home fire deaths. Although the death toll from these fires is much lower than in the 1980s, the rate of death per 1000 reported fires has doubled. Data from the National Fire Incident Reporting System and the National Fire Protection Association's annual Fire Experience Survey were used to create national estimates of these fires and casualties. Including both fires in which upholstered furniture was the item first ignited and contributed most to fire spread, in 2013‐2017, upholstered furniture was involved in an estimated average of 7120 (2%) home fires per year, resulting in an average of 570 (22%) deaths per year. Mattresses or bedding were involved in estimated average of 10 530 (3%) home fires per year, resulting in an average of 370 (14%) deaths annually. In most of these fires and deaths, the item was first ignited. Smoking materials were the leading cause of these fires (20%‐26%) and associated deaths (52%‐54%). Various types of operating equipment and small open flames also play a role. Understanding the causes of these fires is essential to develop new strategies to prevent them.

Big Data and Predictive Analytics in Fire Risk Using Weather Data.

The objective of this article is to study the impact of weather on the damage caused by fire incidents across the United States. The article uses two sets of big data--fire incidents data from the National Fire Incident Reporting System (NFIRS) and weather data from the National Oceanic and Atmospheric Administration (NOAA)-to obtain a single comprehensive data set for prediction and analysis of fire risk. In the article, the loss is referred to as "Total Percent Loss," a metric that is calculated based on the content and property loss incurred by an owner over the total value of content and property. Gradient boosting tree (GBT), a machine learning algorithm, is implemented on the processed data to predict the losses due to fire incidents. An R2 value of 0.933 and mean squared error (MSE) of 124.641 out of 10,000 signify the extent of high predictive accuracy obtained by implementing the GBT model. In addition to this, an excellent predictive performance demonstrated by the GBT model is further validated by a strong fitting between the predicted loss and the actual loss for the test data set, with an R2 value of 0.97. While analyzing the influence of each input variable on the output, it is observed that the state in which a fire incident takes place plays a major role in determining fire risk. This article provides useful insights to fire managers and researchers in the form of a detailed framework of big data and predictive analytics for effective management of firerisk.

Modeling Response Time to Structure Fires

It is important to reduce fire department response times to incidents to improve communities’ general safety, to make the allocation of emergency resources more efficient, and to improve situational awareness. In this article, we identify which factors affect turnout times and travel times for the Arlington County Fire Department in Virginia by applying both linear and spatial models to the U.S. National Fire Incident Reporting System (NFIRS) data. The uniformity of NFIRS data makes this article’s methodological innovations applicable to other participating fire departments in the United States and advances the effort to incorporate scientific evidence into government-level policy-making.

Statistical Analysis of Fire Department Response Times and Effects on Fire Outcomes in the United States

Understanding the relationship between fire department response times and fire outcomes is critical for understanding the impact of a fire department in a community . This paper outlines a statistical analysis of United States fire department response times to fires in 1–2 family residential dwellings between the years 2002–2017. Using data from the National Fire Incident Reporting System (NFIRS), it is shown that a log-normal distribution provides a reasonable approximation of the empirical distribution of response times. Furthermore, this paper investigates the effect of increasing response times on several different measures of fire severity, including the reported fire spread category, the estimated monetary property and contents losses, and the reported flame damage. It was found that all averaged measures of fire severity increase over the interval of 3–13 min of response time, which is primarily due to the increased likelihood of very severe fire incidents at longer response times. An analysis was conducted to assess the effect of meeting the first unit response time benchmarks outlined in NFPA 1710 on the severity of fire outcomes in a community. It was found that department response times were most strongly correlated with the fraction of fires that do “extreme” damage to at least one story and the fraction of the property value lost. It has been proposed that a power law distribution is appropriate for modeling the distribution the burned areas in urban fires. In order to assess whether this proposition is supported by United States data, a Monte-Carlo methodology has been developed to estimate the burned area indirectly from NFIRS data. Finally, several methods are presented for evaluating data consistency and quality from the NFIRS reports. One of the major findings of this study is the quantification of the effect of response time on various measures of fire loss.

A Probabilistic Methodology for Assessing Post-Earthquake Fire Ignition Vulnerability in Residential Buildings

Post-earthquake fire (PEF) ignition events constitute a secondary consequence of an earthquake and may result in the loss of life and substantial property damage, especially in urban areas where the potential for fire spread and conflagration exists. These secondary hazards can cause severe structural and non-structural damage, potentially more significant damage than the direct damage caused by the initial earthquake, and can lead to catastrophic structural failures, devastating economic losses, and casualties. To manage the impact of PEF in urban areas, it is important to identify the potential ignition sources and quantify the vulnerabilities of these ignition sources as a result of earthquake-induced structural damage. The results of such analyses can be used to offer resiliency improvement investments and mitigation strategies in urban areas located in seismically active regions. Most of the previous PEF studies are data-driven, utilizing ignitions reported following recent earthquakes. However, in areas for which historical PEF data are not available, such as the Midwestern United States, a different methodology for developing a PEF model is needed. This paper describes an analytical model for quantifying the vulnerabilities of residential buildings to PEF by estimating the failure of ignition sources upon a probable seismic event. The underlying concept in developing the method is that (1) ignition sources in residential buildings remain unchanged before and after an earthquake, and (2) the total probability of PEF occurrence can be estimated by adjusting the probabilistic fire occurrence data for normal conditions (everyday operation of ignition sources) to account for the effect of the earthquake. This paper’s contribution to state of the art is in developing a new framework for estimating the probability of PEF for areas in which historical PEF data is unavailable. The developed framework uses the likelihood of ignition occurrence during normal condition as a baseline; this baseline is then adjusted using certain key parameters to capture spatial characteristics, ignitability, and potential seismic intensity of the study area to estimate the probability of PEF as a function of projected earthquake characteristics. The model was tested for St. Louis City as a populated area with potential future earthquake hazard because of its proximity to the New Madrid Fault zone. Using the National Fire Incident Reporting System dataset, the frequency of normal condition ignitions was determined as 1.97E−03 ignition per unit per year. Using the proposed PEF model considering PEFs caused by damage to drift and acceleration sensitive equipment and human actions, the projected frequency of PEF was estimated between 2.79E−06 and 2.81E−06 ignitions per household per year. Using this model, and the average number of households between 2010 to 2015, 175,854 households, it was estimated that in the next 50 years, approximately 25 households would experience fires related to probable earthquake events in St. Louis City.

Quantifying Generalized Residential Fire Risk Using Ensemble Fire Models with Survey and Physical Data

Understanding and quantifying property loss fire risk is critical to enabling decision makers in the fire field to make informed decisions. For example, the impact of various decisions made by fire departments on the fire risk in their community, while qualitatively understood by most parties, is not well quantified. Lack of quantification can lead to misdiagnosis and miscommunication on the value of services that result in sub-par resource allocations that negatively impact constituencies. Additionally, the confounding effects between fire department performance and fire risk in publicly available fire data can cause standard regression models to supply counter-intuitive conclusions based upon valid correlations (due to certain critical variables that are difficult to collect and thus unobserved). A methodology is presented that utilizes United States housing, room layout, and fire incident data, as well as experimental heat release rates, material degradation rates, and thermophysical property data in conjunction with physical fire models to estimate a community-averaged extent of fire damage in homes. Housing layout data for single family residential homes were collected from the American Housing Survey. Five home categories representing 45% of U.S. homes were selected for analysis. The U.S. Fire Administration National Fire Incident Reporting System database was used to select the distribution of initial fuels and ignition locations. Survey data was taken to specify furniture layout within the homes. A total of 5167 scenarios were developed for the combinations of home geometry, first item ignited, and furniture layout. Fire evolution was predicted for these scenarios using CFAST and coupled to a pyrolysis-inspired damage model using heat flux to targets in the homes. A damage evolution probability function was constructed for the ensemble of home and furnishing layouts and ignition distributions. This damage model was exercised in a decision analysis problem to demonstrate the utility of the methodology for community-scale resource allocation. Sensitivity studies on the model are likewise performed, indicating the largest source of uncertainty to be linked to the choice of surrogate material properties when calculating home damage. Comparison to best available data is made to assess the robustness of the model.

U.S. Firefighter Injuries on the Fireground, 2010–2014

Firefighting is a hazardous occupation and the fireground is the location for many firefighter injuries, making information about firefighter injury events critical for injury prevention efforts. This study uses data collected through the National Fire Incident Reporting System and the National Fire Protection Association’s annual survey of municipal fire departments to profile U.S. firefighter injuries occurring on the fireground from 2010 to 2014. The research estimates that firefighters experienced an annual average of 30,290 fireground injuries per year in this 5-year period. The findings also indicate that most injuries occurred at residential fires, that one-third resulted in lost work time and that the leading cause of injury was overexertion or strain. Other leading causes of injury were slip/trip, contact with object, falls, and struck or assaulted by person, animal, or object. The study also provides data on primary injury symptoms, place of injury, activity when injured, factors involved in the injury, and injuries by month. Efforts to reduce injury should include providing and requiring the use of appropriate personal protective equipment, establishing and enforcing stringent safety practices, providing firefighters with ongoing health and safety training opportunities, and maintaining fitness, flexibility, and other programs that reduce injury risk.

Underreporting of wildland fires in the US Fire Reporting System NFIRS: California

The absence of a comprehensive and accurate database on fire occurrence means that some proportion of wildfires is not reported. This paper examines wildfires reported in the National Fire Incident Reporting System (NFIRS) database and compares it with the satellite-based Moderate Resolution Imaging Spectroradiometer (MODIS) fire detection data and the CAL FIRE Fire and Resource Assessment Program (FRAP) geodatabase to understand underreporting of wildland fires in the NFIRS system. The paper discusses a series of large wildland fires and uses a Generalised Linear Model to identify the conditions where large wildfires go unreported. This paper shows that the NFIRS database, which is the primary fire database in the US, significantly underreports wildland fires. Evidence from California suggests that 32% of fires within local jurisdictions detected via MODIS satellite were identified in the NFIRS database; thus, there could be a significant number of fires going unreported nationwide. Examining eight large fires in California, only 16% of the structures damaged or destroyed, as reported in CAL FIRE, were reported in NFIRS. Areas with underreporting tend to have a higher level of poverty, higher population density, higher level of people without vehicles, lower income and a higher level of single parents with children under 18, as well as a higher level of people under the age of 18.

Local Fire Department Responses to Fires Involving Automobiles, Buses, and Larger Trucks: 2006-2010 Estimates

Automobile fires account for the majority of vehicle fires and vehicle fire deaths. Fires involving larger trucks resulted in a disproportionate share of vehicle fire losses. Although bus fires are less common, they have a much higher rate of fire based on distance driven. Bus fires have the potential to endanger a larger number of passengers. Any efforts to evaluate the merits of proposed fire safety improvements require an understanding of how many fires and deaths are presently occurring and how many might be prevented with the proposed improvements. Data from the U.S. Fire Administration's National Fire Incident Reporting System (NFIRS) and the National Fire Protection Association's (NFPA's) fire department survey were used to estimate the frequency and associated losses of such fires attended by local U.S. fire departments, and the major factors in these fires and losses. The risk of these fires and of automobile deaths overall and from fires resulting from collision or overturn per billion kilometers driven are also included. The majority of these vehicle fires resulted from mechanical or electrical problems, but three out of every five automobile fire deaths resulted from fires associated with collision or overturn. Language: en

A comparative study assessing factors that influence home fire casualties and fatalities using state fire incident data

The major factors that impact residential fire losses and casualties are relatively well known both on the national level and for any given state. Interestingly, relatively little has been done in comparing fire loss data between states. If state fire loss data are compared, one should be able to identify contributing factors that influence differences in fire losses between states. As an example, it is known that construction standards, regulatory policy, socio-economic factors, etc. all influence the frequency and severity of fires. In this study, subsets of National Fire Incident Reporting System (NFIRS) data for the two largest US states by population are studied in order to identify how areas of origin, items first ignited, and heat sources contribute to the odds of casualties or fatalities occurring in fires, as well as to examine differences between the two states for these three factors. Data on residential home fires from 2006 to 2010 were gathered from the Texas and California NFIRS databases, Texas Fire Incident Reporting System and California All Incident Reporting System, respectively, for this purpose. Examination of the datasets separately using logistic regression models emphasized that fires started in the living room or den, fires in which the item first ignited was a flammable liquid, piping, or filter, and fires that were initiated from cigarettes, pipes, and cigars, all have significantly high odds of resulting in both casualties and fatalities for both states. Additionally, logistic regression modeling with interactions between state and area of origin, item first ignited, and heat source indicated that for many categories, the odds of a fire resulting in a casualty in Texas is roughly 1.5 times higher than the same fire in California. Language: en

Highway vehicle fire data based on the experiences of US fire departments

ABSTRACTIn 2003–2007, US fire departments responded to an average of 267 600 highway vehicle fires per year. These fires caused an average of 441 civilian deaths, 1326 civilian injuries, and $1.0bn (in US dollars) in direct property damage annually. Highway vehicles include cars, trucks, and other vehicles designed for highway use; highway vehicle fires can occur anywhere, not just on a highway. While these fires and associated losses have been falling in recent years, highway vehicles fires accounted for 17% of reported US fires, 12% of US fire deaths, 8% of US civilian fire injuries, and 9% of the direct property damage from reported fires. Data from the US Fire Administration's National Fire Incident Reporting System and the National Fire Protection Association's fire department survey were used to provide details about the circumstances of highway vehicle fires.Mechanical or electrical failures caused roughly three‐quarters of the highway vehicle fires but only 11% of the deaths. Collisions and overturns were factors contributing to the ignition in only 3% of the fires, but fires resulting from these incidents caused 58% of these vehicle fire deaths. The rate of bus fires per billion miles driven was 3.5 times that for highway vehicle fires overall. Copyright © 2012 John Wiley & Sons, Ltd.

Wildland Fires within Municipal Jurisdictions

Each year, wildland fires threaten structures and occupants of the wildland-urban interface (WUI). Currently, wildfire ignition estimates largely exclude ignitions originating within municipal jurisdictions, which contain the majority of the US population. The objective of this article is to provide national estimates and trends of the WUI fire problem; in particular, we analyze the US Fire Administration's National Fire Incident Reporting System fire incident data from 2002 to 2006 reported by local municipal fire departments across the United States. We estimate that, on average, the burning of wildland fuels is associated with 116,971 fires annually. Each year, these fires are responsible for, on average, 15 civilian (nonfire service) fatalities, 88 civilian injuries, and $160 million in direct property losses. These damages include losses to 599 residential structures, 649 nonresidential structures, and 829 vehicles (per year). Based on the value of a statistical life ($8.75 million) and statistical injury ($189 thousand), we find that the economic value of fatalities and injuries averaged $148 million annually. Thus, the losses associated with fires occurring on municipal lands total on average $308 million annually over this time period.

Smoke Alarm Presence and Performance in U.S. Home Fires

The vast majority of U.S. homes have at least one smoke alarm. For smoke alarms to be effective, they must have a functional power source, be close enough to the smoke to activate, they must be heard, and occupants must take appropriate action. In homes with smoke alarms and fires considered large enough, the alarms operated 83% of the time. Analyses of data from the U.S. Fire Administration’s National Fire Incident Reporting System and the National Fire Protection Association’s fire department survey showed that in 2003 to 2006, no smoke alarms were present in 31% of reported home fires and 40% of home fire deaths. Smoke alarms were present but failed to operate in 9% of the reported fires and 23% of the deaths. Thirty-seven percent of the deaths resulted from the 47% of fires with operating smoke alarms. Circumstances of the fire, the detection equipment, and occupant characteristics must be considered when evaluating smoke alarm performance. Hardwired smoke alarms operated more often than alarms powered by batteries alone. Victims of fatal fires with working smoke alarms were more likely than victims of fires without working smoke alarms to have been in the area of origin when the fire began; fighting the fire; unable to act; or at least 65 years old; and less likely to have been sleeping. Findings from the Consumer Product Safety Commission’s 2004 to 2005 Residential Fire Survey provide background about smoke alarms in the general population and in unreported fires.