We studied the foraging behaviour and energetics of the nectar-feeding bat, Leptonycteris curasoae (=L. sanborni), in the Sonoran Desert near Bahia Kino, Sonora, Mexico, using radio-telemetry, light-tagging, and focal plant observations to answer three questions: (1) How far do these bats fly in a night and at what energetic cost? (2) How do they harvest nectar and pollen from columnar cacti that offer large but temporally variable nectar rewards? (3) What are the implications of their foraging behaviour for gene flow within populations of their food plants? L. curasoae visited flowers of three species of columnar cacti in April through June. Many bats roosted on Isla Tiburon 20 km from the Mexican mainland and commuted about 30 km to the mainland to feed. Bats flew for about 5 h each night for a total distance of about 100 km. Individuals foraged alone or in small groups in overlapping areas of 1-3 km2; and visited the same feeding areas on successive nights. Within feeding areas, bats visited the flowers of many cactus plants, visited most flowers <5 times, and removed about 0.1 mL of nectar per visit. Although bats flew nearly continuously early in the evening, they did most of their feeding between 24:00 and 02:00. When visiting flowers of Pachycereus pringlei (cardon), bats apparently waited until flowers had accumulated 0.8 mL of nectar before feeding, which suggests that rates of nectar secretion influence the timing of feeding in these bats. We estimate that the daily energy budget of L. curasoae is at least 40 kJ and that bats make about 80-100 visits to cactus flowers to acquire this energy. Foraging areas typically contain thousands of cactus flowers, and thus food does not appear to be a limited resource for these bats during the spring. The cost-efficient flight of this bat makes it an excellent pollen vector for self-incompatible, widely spaced desert cacti.