Many orchid species are unusual in that they provide no nectar or pollen rewards for their pollinators. Absence of reward is expected to have a fundamental effect on pollinator visitation patterns. In particular the number of flowers visited per inflorescence is expected to be affected in both unrewarding and co-flowering rewarding species. We used arrays of artificial inflorescences, which could be either rewarding or unrewarding and were differentiated by their colour, to test how many flowers bumblebees visit in each type of inflorescence. The frequency of the two colours was varied, thus modelling the case where different frequencies of both an unrewarding and rewarding species were present in a patch. We found that bumblebees visited more flowers per rewarding inflorescence after they have experienced unrewarding or partially emptied rewarding inflorescences. We used these results to simulate pollen transfer and thus predict selfing rates on rewarding inflorescences. We found these increased when nectar depleted or when there was a greater proportion of unrewarding inflorescences in the patch. Conversely, we found that the number of flowers bumblebees visited on each unrewarding inflorescence did not significantly change through experiments. Selfing rates for unrewarding inflorescences were predicted to depend principally on the number of these inflorescences bumblebees visited rather than on the number of flowers they visit per inflorescence. This was because most visitors to orchids are supposed to be naive, and pollinators that commence foraging carrying no pollen will necessarily self any flower they pollinate on the first inflorescence they visit. Thus the average selfing rate is expected to increase as the sequence of inflorescences visited decreases in length.