Abstract The objective of this study was to determine how closely acute (72-hr) lethal concentration estimates developed from probit analysis compared to the demographic toxicological endpoints, net reproductive rate ( R o ), the intrinsic rate of increase ( r m ), and realized fecundity ( U x ), in terms of predicting effects of pesticides on populations. Lethal and sublethal effects of the insecticide imidacloprid on the arthropod Acyrthosiphon pisumHarris (pea aphid) were determined for populations exposed to foliar-sprayed broad bean Vicia fabaL. (variety Banner). An examination of R o indicated that sublethal effects were occurring that reduced reproduction. However, by looking at the mean number of offspring produced per surviving female and U x , it was determined that the reduction in R o was entirely due to acute mortality and a reduction in life span. Also, exposure to increasing concentrations of imidacloprid did not cause a shift in either the day of initial reproduction or the day of peak reproduction. Therefore, this pesticide caused no sublethal effects on reproduction and, as such, a lethal concentration estimate should have been a good predictor of effect at the population level. However, the 72-hr lethal concentration estimate was not a good predictor of effect of this pesticide on population growth. Populations exposed to the 72-hr LC 60were able to maintain rates of population increase ( r m = 0.224) similar to those of the control ( r m = 0.295). The data indicate that the reason for the discrepancy between acute lethal concentration estimates and population growth was that surviving individuals were able to sustain heightened rates of reproduction following acute exposure to imidacloprid. The ability of surviving individuals to maintain these high reproductive rates allowed them to compensate for losses and act as reservoirs for future reproduction. It is not possible, using acute mortality estimates alone, to predict this “reservoir effect,” and therefore not possible to predict how a population's growth rate will respond or change based on this endpoint. Thus this would suggest that the assessment of a xenobiotic based solely on acute mortality estimates will lead to flawed conclusions about a population's exposure response.