Aerodynamic theory predicts that minimum power (Vmp) and maximum range (Vmr) flight speeds increase when the body mass of an individual animal increases. To evaluate whether foraging bats regulate their flight speed within a fixed speed category relative to Vmp or Vmr, I investigated how the natural daily changes in body mass caused by feeding affected the flight speed of neotropical nectar-feeding bats (Phyllostomidae: Glossophaginae) within a strictly defined, stereotyped behavioural context. Individual bats were maintained in a flight tunnel (lengths of five different types 14-50 m) with a fully automated feeding, weighing (using an electronic balance at the roost) and flight speed measuring system. Flight speeds were measured during normal nocturnal foraging activity by an undisturbed bat while it flew between the two ends of the flight tunnel to obtain food from two computer-controlled nectar-feeders. For a comparison of flight enclosure measurements with field data, flight speeds were also obtained from unrestrained bats foraging in their natural environment (Costa Rica). Foraging flight speeds spanned a range of at least a factor 3 within a single species, which demonstrates the wide range of speeds possible to these animals. Significant, positive correlations between flight speed and the natural individual variability in body mass were found in nearly all cases, with body mass exponents ranging between 0.44 and 2.1. Bats flying at normal speeds were therefore not near their upper limit of muscle power. The most reliable measurements of speed increase with mass (with individual mass changes of up to 30%) were close to the increase theoretically predicted for Vmp and Vmr for an individual bat (with constant wing span and area), which should vary as M0.42, where M is mass. This provides evidence that the glossophagine bats attemped to maintain their flight speed within a fixed speed category relative to Vmp or Vmr during foraging. Among differently sized species of glossophagine bat (N=4), flight speeds V varied with V=20M0.23, in agreement with the mass exponent of 0.21 expected from aerodynamic models for interspecific variation. In addition to the mass effect, at least five other variables significantly influenced flight speed. (1) Both mean and maximum flight speeds increased with the length and the cross-sectional area of the flight tunnel. Mean (maximum) flight speeds of 11-12 g Glossophaga soricina bats (in m s-1) were 4.6 (5.3) over a 7 m and 7. 3 (10.5) over a 50 m flight path. (2) The flight speed range adopted by a bat during one night could vary significantly between nights, independently of body mass and the effect of the size of the flight enclosure. (3) Bats flew significantly faster under illumination than in darkness. This effect was shown (i) by bats kept under natural ambient illumination that initiated foraging during the twilight phase of the evening, (ii) when bats continued to feed into the light phase directly after the dark-light transition in the laboratory and (iii) during foraging under constant, artificial illumination. (4) After a period of rest, the initial flight speed during a foraging bout was significantly increased by 25%, but declined to the mean level within 20 s of activity. (5) Flight speed could differ significantly between foraging (flight from feeder to feeder) versus non-foraging (flight from end to end of the enclosure without visiting the feeders) flights. The results of this study demonstrate a clear ability of bats to regulate their flight speed in response to small natural changes in body mass as predicted by aerodynamic theory for Vmp and Vmr. The set point in flight speed regulation, however, was influenced by multiple additional variables.