The delayed alternation behavior of rats was investigated in four experiments with a T-maze. Each trial involved an initial forcing of the subject to one arm of the T, followed either 0 or 40 sec later by a free-choice run on which a turn in either direction could be made. In order to obtain a reinforcer on the free-choice run, the rat was required to turn in the direction opposite that of the forcing. During experimental testing, delayed alternation target trials were preceded by an initial delayed alternation trial which was either the same as or opposite to the target trial. Relative to a control in which an initial trial was not presented, accuracy on the free-choice run of target trials was reduced by either type of initial trial (interference effect). Experiments 1 and 2 demonstrated that (a) interference was obtained when the target-trial retention interval was 40 sec but not when it was 0 sec and (b) interference was reduced by repetition of the initial trial and by increased temporal separation between the initial and the target trials. Experiment 3 demonstrated that an initial trial that consisted entirely of forced runs produced an interference effect at both the 0- and 40-sec retention intervals. The findings are discussed in relation to previous studies of interference effects in rats and pigeons.