Abstract Most plants require animal pollination to reproduce, prompting concern that pollinator declines immediately threaten plant populations. This concern is warranted if pollinator-mediated seed losses cause declines in plant population growth rates (λ). However, demographic trade-offs might reduce the risk of population decline if seed loss improves performance elsewhere in the life cycle. We conducted a multi-year pollination manipulation on four species and measured how demographic vital rates and λ responded. Seed responses did not predict net changes in λ. Reduced pollination decreased seed production, but only caused a net decrease in λ in one species; in the others, improved survival buffered λ. Increased pollination boosted seed production, but at a cost to survival that caused a net reduction in λ in three species. Our results highlight the importance of demographic trade-offs for understanding the impacts of pollinator declines on plant biodiversity and, more broadly, the population-level impacts of changing mutualisms.