Quantifying the benefits of pollinator abundance and diversity for plant reproduction is a critical challenge, especially given recent pollinator declines. Many studies test for pollen limitation (PL), but huge gaps remain in our understanding of how often and why plants experience PL. Most studies are limited to a few sites or years. Little existing research mechanistically links pollinator visitation with plant reproduction. Also, buffering against variability in visitation is a key but largely untested potential benefit of pollinator diversity. Long‐term studies pairing visitation data with PL experiments are essential to addressing these gaps, but extremely rare. I measured pollinator visitation, responses to pollen supplementation, and fruit weights in the generalist plant Scrophularia californica across multiple sites for 8 years. I determined how often S. californica experienced significant PL and tested mechanistic models predicting PL and fruit (capsule) weight from visitation. I then used those models to simulate pollinator loss and evaluate the relative effects of changes in visit quantity and quality. I explored the stabilizing benefits of visitor diversity by quantifying both covariation among pollinators and effects of simulated species loss on variation in PL and capsule weights. Scrophularia showed significant PL in only 26.9% of patch/year combinations tested. Visitation by the most abundant (Apis mellifera) and effective (Vespula pensylvanica) pollinators strongly predicted PL and capsule weights. Both response curves were highly nonlinear, but PL saturated more with increased visitation. Pollinator loss simulations predicted substantially increased PL and reduced capsule weights, mostly because of lower visit numbers rather than changes in visit quality. Visitation rates of individual pollinators varied substantially and without correlation. Simulated pollinator loss increased the spatial and temporal skew in visitation, resulting in higher predicted coefficients of variation for PL and capsule weights. Synthesis. Previous work predicts a nonlinear relationship between pollinator visitation and plant reproduction, but this is the first long‐term study I am aware of that quantifies such a pattern. The analysis of covariance in pollinator visitation combined with my simulation results provide novel empirical support for another important hypothesis, that multiple pollinators can provide stabilizing benefits for plant reproduction.