A study was conducted to determine the influence of initial population density on the effects of pesticides on pea aphid, Acyrthosiphon pisum (Harris), populations. Three initial starting densities of pea aphids (147, 295 and 590 aphids per m(2)) were exposed to no pesticide or imidacloprid at rates of 1 or 5 g ai/ha on broad bean plants, Vicia faba L., in a greenhouse. Ten days later, population size was assessed. In general, higher initial aphid population density resulted in a higher final population density for all imidacloprid concentrations. However, population growth rates for populations started with the highest density (590 aphids per m(2)) were significantly lower than those with initial densities of 147 and 295 aphids per m(2). This was due to a relative reduction in population number. Populations begun with 147 aphids per m(2) were 50% lower after exposure to the highest concentration of imidacloprid, whereas the populations begun with 295 and 590 aphids per m(2) were 42 and 25% of the starting population size, respectively. Therefore, the pesticide actually had a greater impact on the population started with the highest density. This can be explained by a synergistic effect of the pesticide and crowding. The lower growth rate observed in the population started with the highest density was probably due to crowding, whereby aphids approached the carrying capacity and were stressed. Even though these populations were reduced, final density was still sufficiently high to limit resources. These results indicate that the response of organisms to stress is influenced by population density at the start of a stressful event, such as a pesticide exposure. Therefore, different experimental designs may result in different outcomes and starting population densities must be carefully considered when designing population-level toxicological experiments.