Abstract Over 50% of dogs in the USA are overweight or obese indicating that owners find it difficult to know how much food their dog needs. Feeding guidelines are based on the average energy requirements of dogs and do not take into account measured activity levels, which may lead to overfeeding by well-intentioned owners. The most common published energy prediction equation is (RER*AF) where RER is Resting Energy Requirement [70*(BW kg)0.75)] and AF is the Activity Factor, which generally ranges between 1 (body weight loss) and 8 (heavy work). This equation requires the user to estimate activity levels when selecting an AF, which may result in under- or overfeeding. The objective of this pilot study was to determine the ability of collar-worn activity monitors to more accurately predict the energy expenditure of dogs. Ten weight-stable beagles were group-housed at the Hill’s Pet Nutrition Center in Topeka, KS. Dogs were provided daily access to open, outdoor play space and human interaction; water was freely available. Dogs were weighed before feeding on days 1 and 5 of each of the 2 test periods. Accelerometer-based activity monitors were attached to the collars of the dogs on day 1 of each test period and activity was continually measured throughout the study using the activity monitor. Daily food consumption in grams was recorded and was converted to energy intake using the caloric density of the food. Daily energy requirements were calculated for each dog using the standard RER equation and either a standard AF of 1.6 or the estimated energy required for activity, in which newly established algorithms for quantification of walking, trotting, and running behaviors were paired with estimated caloric expenditure based on level of exertion (quantified health approach). The mean absolute error (MAE) of each method was calculated as the absolute value of the difference between the daily energy requirement vs. the daily caloric intake for each dog. The MAE was averaged across all dogs and all days for each of the two methods. The MAE ± SD of the standard RER and AF was 103 ± 126 kcal/day (Range -21.5 to 216 kcal/day). The MAE ± SD of the quantified health approach was 86 ± 90 kcal/day (Range -37.6 to 164.4 kcal/day). Both approaches resulted in normal MAE distributions. In this pilot study, quantifying energy expenditure using a wearable activity monitor reduced over-feeding by 16.5% compared with standard equations. In the future, energy predictions based on measured activity could result in more accurate feeding recommendations than traditional approaches. Future research aimed at optimizing energy predictions may help address the problem of pet obesity. This study was approved by the Hill’s Internal Animal Care and Use Committee. Funding: This study was funded by Hill’s Pet Nutrition, Inc.