Climate strongly influences the population dynamics of many species, but intrinsic and extrinsic factors such as density-dependence and anthropogenic impacts can confound the effects of climate. Further, the temporal scale of climate response is determined by the unique characteristics of a species' life history, and determining the most appropriate climate indicator at the proper scale is a challenge faced by population ecologists. We focused on how climate influences juvenile survival of bocaccio (Sebastes paucispinis), a threatened Pacific rockfish, because its abundance has declined >90% in the last 25 years, ostensibly as the result of overfishing. Bocaccio recruitment is episodic, with strong recruitment events apparently related to climate conditions. We developed a sequence of models that related log of juvenile survival to the predictor variables population density and climate, as measured by the Northern Oscillation Index. A model that contained only population density as a predictor variable explained only 1.4% of the variance, while a model that included only climate indices explained 52%. Including density additively with climate did not improve model fit. However, a model that included an interaction between density and climate explained more than 68% of the variance. In addition, models that represented climate as monthly indices fit the juvenile survival data much better than those that averaged climate over 2-or 3-month periods. Our results suggest that climate affects bocaccio recruitment as a series of pulses corresponding to particular life-history events, with population density mediating the magnitude of the climate effect during the settlement stage.