For organisms with restricted mobility, density dependence may occur on spatial scales much smaller than that of the whole population. Averaging densities over whole populations in such organisms gives a more or less inaccurate description of the real variation in competitive intensity over time and space. The potential for local density dependence in more mobile organisms is less well understood, particularly for organisms living in continuous habitats. To test for local density-dependent processes in such an organism, we manipulated egg density (the number of eggs nest(-1)) among ten artificial nests of Atlantic salmon along an 1,848-m long river during two consecutive years. Eggs in different nests were given unique thermal otolith-banding patterns to allow identification of juvenile nest origin. At capture, 1-2 months after emergence, the spatial distribution of juveniles reflected nest locations, with the median absolute dispersal distance being 92 and 41 m in the 2 years. Estimated nest-specific survival rates were strongly negatively related to hatched-egg density in both years (r(2)=0.72 and 0.62), despite dramatic differences in overall mean survival (0.22 and 0.02). Thus, density-dependent survival following emergence in Atlantic salmon juveniles occurs on spatial scales much smaller than that of whole populations. The consistency across years suggests that the phenomenon is likely to occur over most environmental conditions. Our observation of local-scale density dependence is consistent with strong juvenile territoriality, which forces individuals emerging in high-initial density areas to disperse farther, and a high cost (metabolic or predation) of dispersal. We conclude that for mobile organisms with patchy distributions of propagules and constrained juvenile dispersal, increased emphasis on local-scale dynamics should enable a more mechanistic understanding of population regulation even in continuous habitats, and hence increase the predictive power of population models.