When offered Aedes aegypti larvae of two different sizes, P. nymphula larvae showed no frequency-independent or frequency-dependent prey preference. The two-prey version of D. ROGERS’(1972, J. Anim. Ecol. 41: 369-383) random predator equation predicted accurately the proportion of each prey type taken by the dragonflies; an optimal diet model did not. These results are contrasted with those obtained in similar experiments on Enallagma cyathigerum (S.H. CHOWDHURY et al„ 1989. Odonalologica 18: 1-13).