A mechanism is described which is incorporated into a pitfall trap to separate the catch into 6 periods of activity during the day. An experiment to determine the bias of the trap showed that underestimates are made of the number of oribatid mites and parasitic Hymenoptera. Of 17 groups into which catch was classified, these were only 2 giving wrong estimates. Pitfall traps are shown to be sufficiently reliable, in the statistical sense, to reveal differences in number of animals caught at different times of the year, and in different habitats.