Lake Superior rainbow smelt (Osmerus mordax) recruitment to 12-13 months of age in the Wisconsin waters of Lake Superior varied by a factor of 9.3 during 1978-1997. Management agencies have sought models that accurately predict recruitment, but no satisfactory models had previously been developed. In this study, modeling was conducted to determine which factors best explained recruitment variability. The Ricker stock-recruitment model derived from only the paired stock and recruit data accounted for 63% of the variability in recruitment data. The functional relationship that accounted for the greatest amount of recruitment variation (81%) included rainbow smelt stock size, May rainfall, and bloater (Coregonus hoyi) biomass. Model results were interpreted to mean that recruitment was affected negatively by increased river flows from increased rainfall, and affected positively by the biomass of bloater, and those results were interpreted to mean that bloater mediated the effects of lake trout predation on rainbow smelt recruits. Model results were also interpreted to mean that stock size caused compensatory, density-dependent mortality on rainbow smelt recruits. Correlations observed here may be of value to managers seeking approaches to either enhance or control populations of this species, which is not indigenous to the Great Lakes.