Ecological risk assessments rarely evaluate indirect pesticide effects. Pesticides causing no direct mortality in wildlife can still reduce prey availability, resulting in a lower reproductive rate or poor juvenile condition. Few studies have examined these consequences at the population level. We use a four-year data set from a forest ecosystem in which Bacillus thuringiensis kurstaki (Btk) was applied to control gypsy moths (Lymantria dispar L.). Lower worm-eating warbler (Helmitheros vermivorus) productivity on Btk plots contributed to an intrinsic growth rate <1. Altered provisioning behavior by adults led to lower nestling mass in Btk-treated plots, and simulations of reduced juvenile survival expected as a result further reduced population growth rate. The present study explored different spatial representations of treated areas, using a two-patch matrix model incorporating dispersal. Minimal migration from areas with increasing subpopulations could compensate for detrimental reductions in reproductive success and juvenile survival within treated subpopulations. We also simulated population dynamics with different proportions of treated areas to inform management strategies in similar systems. Nontoxic insecticides are capable of impacting nontarget populations with consistent, long-term use and should be evaluated based on the spatial connectivity representative of habitat availability and the time period appropriate for risk assessment of pesticide effects in wildlife populations.