Failure to account for abiotic factors in assessing human effect on raptor behavior can result in dubious conclusions because the same data can produce opposing management recommendations depending on use of weather variables as covariates in analyses. We illustrate this problem with examples from a study of the potential effects of military training on raptor behavior. Variation in weather explained more (P 0.001) variation in raptor abundance than did military activity. Decreasing residual error, by controlling for weather variation in analyses, reversed conclusions drawn from 2 of 4 tests of military influence on raptor abundance. Pooling data reduced the need for controlling weather factors by increasing test power to detect small differences, but sacrificed the ability to detect temporal effects of training, which might be useful to managers. Blocking data by season and time of day did not control for variation in weather; when weather factors were included as covariates, conclusions were reversed in 2 of 9 temporal/seasonal blocks in each of 2 analyses testing the effect of military training on prairie falcon (Falco mexicanus) occurrence. Weather data obtained at the study site during observations explained more (P = 0.009) variation in raptor abundance than did data obtained from the National Weather Service. However, selecting weather data from either source resulted in similar management conclusions. Researchers must address environmental factors that are likely to influence animal behavior but that cannot be removed from the study design.