The relationship between growth rate and life span was studied in Drosophila by varying the amount of yeast available to each developing larva at constant temperature, 25 degrees C. With one approach the larvae developed in a standard medium at constant larval density and a varying amount of yeast added on the medium. Across the entire growth rate range covered in this way (10-100 micrograms/day, male flies) imaginal life span depended on growth rate in a biphasic way, the relationship having a parabolic form with a maximum at about 55 to 60 micrograms/day. Similar covariation of growth rate and life span was obtained by varying larval density at a constant amount of added yeast. With both these approaches growth rate variation was due to opposite variations of both components of growth rate, i.e. duration of development and body size. However, development in a medium without nutrients but with a varying amount of added yeast at constant larval density led to a similar biphasic relationship between growth rate and life span although duration of development did not vary. Therefore, the present results are not compatible with the hypothesis that there is a single causal negative relationship between growth rate and life span and demonstrate that duration of development is not a causative factor of the biphasic relationship between growth rate and life span established here.