The design and construction of 2 multiple Y-unit mazes are described, which will permit the assessment of the mean and variance of phototactic behavior in Drosophila populations. Using maze performances as criteria, selection has been applied for 15 generations. By this procedure highly photopositive and photonegative strains have been produced. The strength and limits of selection in the different mazes are established. By an analysis of the behavior of the selected and unselected strains, the interaction of the environmental and genetic influences of phototactic behavior in Drosophila melanogaster is demonstrated. The necessity of recognizing individual differences in populations of experimental animals and the importance of a controlled environment in the study of photo-taxis are discussed, with particular reference to Drosophila.