Pollinator abundance is declining worldwide and may lower the quantity and quality of pollination services to flowering plant populations. Loss of an important pollinator is often assumed to reduce the amount of pollen received by stigmas of a focal species (pollination success), yet this assumption has rarely been tested experimentally. The magnitude of the effect, if any, may depend on the relative efficiency of the remaining pollinators, and on whether the loss of one pollinator leads to changes in visitation patterns by other pollinators. To explore how a change in pollinator composition influences pollination of Asclepias verticillata, we excluded bumble bees from plots in large and small populations of this milkweed species. We then quantified pollinator visitation rates, pollen export, and pollen receipt for control plots and for plots where bumble bees were experimentally excluded. We found that exclusion of bumble bees did not reduce pollen receipt by A. verticillata flowers. Visitation by Polistes wasps increased markedly following bumble bee exclusion, both in small populations (186% increase) and in large populations (400% increase). Because Polistes wasps were as efficient as bumble bees at pollen transfer, increased wasp visitation offset lost bumble bee pollination services. Thus, loss of a frequent pollinator will not necessarily lead to a decline in pollination success. When pollinator loss is followed by a shift in the composition and abundance of remaining pollinators, pollination success will depend on the net change in the quantity and quality of pollination services.