The locomotor activity patterns of wildtype Drosophila melanogaster and the mutants so (sine oculis) and sol (small optic lobes) were investigated. In all strains the proportions showing circadian rhythmicity, arrhythmicity and more complex patterns which could not be thus classified were similar. The occurrence of abnormal activity patterns is thus not a property of the mutation as previously claimed (Mack & Engelmann, 1981). In flies with a distinct circadian rhythmicity, the period lengths (τ) varied between strains, τ for wildtype Italy and the mutant so was longer than for wildtype Berlin and the mutant sol. As different τ's have been reported by others, τ does not seem to be closely determined for Drosophila melanogaster. Many flies exhibited two rhythms simultaneously, one with τ shorter and one with τ longer than 24 h, apparently implying two-oscillator control of the locomotor activity. The eyeless so mutants were entrained by LD cycles, so the compound eyes are not necessary, and so must possess the relevant photoreceptor(s) elsewhere. This may therefore also be true for the wildtype. Histology of the so adults revealed no correlation between the degree of reduction in the medulla and the occurrence of abnormal activity patterns. Since the only structures common to the medulla of so and sol are known to be large tangential cells, it is concluded that either they are of importance for the rhythmic system, or the oscillator(s) controlling locomotor activity is (are) not located in the medulla.