1. Spatial heterogeneity in population density is predicted to have important effects on population characteristics, such as competition intensity and carrying capacity. Patchy breeding distributions will tend to increase spatial heterogeneity in population density, whereas dispersal from breeding patches will tend to decrease it. The potential for dispersal to homogenize densities is likely to differ both among organisms (e.g. plants vs. mobile animals) and throughout ontogeny (e.g. larvae vs. adults). However, for mobile organisms, experimental studies of the importance of breeding distributions from the wild are largely lacking. 2. In the present study, experimental manipulations replicated over eight natural streams and 2 years enabled us to test for effects of the distribution of Atlantic salmon eggs over spatial scales which are relevant to local interactions among individuals. Artificial nests were placed along 250 m study reaches at one of two levels of nest dispersion - patchy (two nests per stream) and dispersed (10 nests per stream) - while holding total egg density (eggs m(-2) stream area) constant. 3. Nest dispersion had significant effects on the spatial distribution of the resulting juveniles in their first summer. Patchy nest distributions resulted in a highly right-skewed frequency distribution of local under-yearling densities (among 25 m sampling sections), as sample sections adjacent to the nest sites had relatively high densities. In contrast, dispersed nest distributions yielded approximately normal density distributions. Sections with high relative densities in the patchy nest distribution treatments also had relatively small juvenile body sizes, and patchy egg distribution appeared to produce a higher redistribution of individuals from the first to the second juvenile growth season than the dispersed distribution. 4. Because patchy breeding distribution combined with limited early dispersal can create spatial variation in density over scales directly relevant for individual interactions, this will be one important component in determining mean levels of early juvenile competition and its spatial variation within populations. Assuming random or ideal-free distribution of individuals may therefore underestimate the mean level of density experienced by juveniles over surprisingly small spatial scales (orders of magnitude smaller than total spatial extent of populations), even for mobile organisms.