Objectives: Modeling crash risk in urban areas is more complicated than in rural areas due to the complexity of the environment and the difficulty obtaining data to fully characterize the road and surrounding environment. Knowledge of factors that impact crash risk and severity in urban areas can be used for countermeasure development and the design of risk assessment tools for practitioners. This research aimed to identify the characteristics of the road and roadside, surrounding environment, and sociodemographic factors associated with single-vehicle crash (SVC) frequency and severity in complex urban environments, namely, strip shopping center road segments.Methods: A comprehensive evidence-based list of data required for measuring the influence of the road, roadside, and other factors on crash risk was developed. The data included a broader range of factors than those traditionally considered in accident prediction models. One hundred and forty-two strip shopping segments located on arterial roads in metropolitan Melbourne, Australia, were identified. Police-reported casualty data were used to determine how many SVC occurred on the segments between 2005 and 2009. Data describing segment characteristics were collected from a diverse range of sources; for example, administrative government databases (traffic volume, speed limit, pavement condition, sociodemographic data, liquor licensing), detailed maps, on-line image sources, and digital images of arterial roads collected for the Victorian state road authority. Regression models for count data were used to identify factors associated with SVC frequency. Logistic regression was used to determine factors associated with serious and fatal outcomes.Results: One hundred and seventy SVC occurred on the 142 selected road segments during the 5-year study period. A range of factors including traffic exposure, road cross section (curves, presence of median), road type, requirement for sharing the road with other vehicle types (trams and bicycles), roadside poles, and local amenities were associated with SVC frequency. A different set of risk factors was associated with the odds of a crash leading to a severe outcome: segment length, road cross section (curves, carriageway width), pavement condition, local amenities and vehicle, and driver factors. The presence of curves was the only factor associated with both SVC frequency and severity.Conclusions: A range of risk factors were associated with SVC frequency and severity in complex urban areas (metropolitan shopping strips), including traditionally studied characteristics such as traffic density and road design but also less commonly studied characteristics such as local amenities. Future behavioral research is needed to further investigate how and why these factors change the risk and severity of crashes before effective countermeasures can be developed.