Three groups of rats (n = 4) received runway training in Phase 1 to establish, in each group, one stimulus as a signal for a single nonreinforced (N) trial and another stimulus as a signal for four consecutive N trials. All groups received consistent reinforcement (CR) in Phase 2, with Group N-4 receiving the signal for four N trials and Group N-1 receiving the signal for one N trial, twice each day. Group CR received no signal presentations in Phase 2. Phase 2 was followed by 20 unsignaled extinction trials, 5 trials per day. As predicted by the signal-generated memory hypothesis, Group N-4 was more resistant to extinction than Groups N-1 and CR.