We used multiple linear regressions across a Northeast Pacific region 56-lake set to examine the influence of sockeye salmon spawner densities and limnological, climatic, and watershed characteristics on sediment stable nitrogen isotope (δ15N) and C:N signatures, geochemical proxies used in paleolimnological reconstructions of prehistoric salmon abundances. Across all sites (n = 56), sedimentary C:N was primarily related to variables reflecting allochthonous organic matter and nutrient fluxes, with 67% of the variance explained by watershed vegetation type, watershed area, mean slope, and salmon spawner densities. In a subset of sites with nutrient data (n = 41), 63% of variance in C: N was explained by precipitation and watershed vegetation type. Sediment <515N was most strongly related to salmon spawner densities, but C:N and watershed area: lake area ratios explained significant residual variance (total 55%). In sites with nutrient data, salmon spawner densities, vegetation type, and spring nitrate explained 62% of the variance in δ15N. Sediment C: N and δ15N values exhibited distinct interregional variation, typically varying inversely. Regional δ15N-salmon density relationships (regression slopes) varied strongly with total annual precipitation (r2 = 0.89, p = 0.016, n = 5), suggesting watershed organic matter and nutrient loading vary predictably in relation to regional biogeoclimatic conditions. Our findings demonstrate that paleolimnological analyses are useful for quantitative reconstruction of past salmon densities; however, inferences regarding past salmon populations must consider the factors regulating influxes of nitrogen from watershed sources.