Departure rules used by solitary long-tongued bees (Anthophora spp. andEucera spp.) collecting nectar from flowers ofAnchusa strigosa (Boraginaceae) were studied. The amount of nectar a bee receives from an individual flower was estimated by measuring the time elapsed since the previous bee visit to that flower. Measurements of nectar accumulation in experimentally emptied flowers indicated that this time interval is an accurate predictor of nectar volumes in flowers. We found that nectar rewards influence the probability of departure from individual plants, as well as distances of movements within plants. The probability of departure from individual plants was negatively related to the amount of reward received at the two lastvisited flowers. This result indicates that the bees used a probabllistic departure rule, rather than a simple threshold departure rule, and that rewards from both the current and the previously visited flower were important in determining departure points. Distances of inter-flower movements within plants were negatively related to the amount of reward received at the current flower. The overall results suggest that the pollinators ofA. strigosa make two types of departure decisions-departures from the whole plant and departures from the neighbourhood of individual flowers-and that they use different departure rules for each scale. Factors influencing the decision-making processes of the observed foraging behaviour are discussed.