The partial reinforcement extinction effect (PREE) is a well-known phenomenon in associative learning. It refers to the paradoxical observation that after an animal has been trained on a partial reinforcement (PRf) schedule, it exhibits greater responding in the face of extinction than an animal that was trained on a continuous reinforcement (CRf) schedule, even if it has received fewer pairings of the conditioned stimulus (CS) and the unconditioned stimulus (US). The current thesis investigated the mechanisms that produce the PREE. We showed that the PREE is eliminated when partially reinforced rats are extinguished using more trials, but not when they are extinguished using longer (but the same number of) trials. This showed that the PREE is influenced primarily by the number of trials presented during extinction. In addition, we found that the number of trials taken to extinguish responding is directly proportional to the proportion of trials reinforced during conditioning. We then investigated accounts of the PREE which suggest that the generalisation of responding during nonreinforcement from partial reinforcement to extinction produces the PREE. There was no effect of pre-exposure to nonreinforcement on the rate of acquisition to a PRf schedule and a CRf schedule, indicating that the process that produces the PREE is not symmetrical as would be expected from a generalisation account. Finally, the role of nonreinforced trials in producing extinction was examined by extinguishing responding while maintaining reinforcement at a much lower rate than during conditioning. We found a PREE despite continuing reinforcement, indicating that animals learn about both the per-trial outcome of the schedule and the expected time to reinforcement. Our results present a challenge for trial-based and time-based models of extinction and the PREE.