Recruitment establishes the initial size of populations and may influence subsequent population dynamics. Although strong inference can be made from empirical relationships between recruitment and population sizes, a definitive test of recruitment limitation requires manipulating recruitment at relevant spatial and temporal scales. We manipulated oviposition of the mayfly Baetis bicaudatus in multiple streams and measured the abundance of late-stage larvae at the end of the cohort. Based on fundamental knowledge of mayfly behavior, we increased, eliminated, or left unmodified preferred mayfly oviposition sites in 45-m reaches of streams (N = 4) of one high-altitude drainage basin in western Colorado, USA. We compared egg densities before (2001) and after the manipulation (2002) using paired t tests and compared larval densities before and after the manipulation among treatments using repeated measures analysis of variance. This manipulation altered not only egg densities, but also larval abundances 1 year later. Compared to the previous year, we experimentally increased egg densities at the addition sites by approximately fourfold, reduced egg densities to zero in the subtraction sites, and maintained egg densities in the control sites. After the manipulation, larval densities increased significantly by a factor of approximately 2.0 in the addition sites and decreased by a factor of approximately 2.5 in the subtraction sites. This outcome demonstrates that dramatic changes in recruitment can limit larval population size at the scale of a stream reach, potentially masking previously observed post-recruitment processes explaining the patterns of variation in abundance of a stream insect. Furthermore, our results emphasize the importance of preferred oviposition habitats to population sizes of organisms.