Three experiments with male and female rats were conducted to examine the effects of Pavlovian extinction training on Pavlovian-to-instrumental transfer (PIT) in a task in which the unconditioned stimulus (US) was presented at an early time point within an extended conditioned stimulus (CS). Two instrumental responses were trained with different reinforcing outcomes (R1-O1, R2-O2) and then, independently, 2 stimuli were trained with those outcomes (S1-O1, S2-O2). One group then underwent an extinction treatment (S1-, S2-) and a second was merely exposed to the experimental contexts without any stimulus events. Finally, the effects of the 2 stimuli on instrumental responding were assessed in PIT tests. Across experiments we varied the number of Pavlovian training trials prior to extinction (8, 16, or 64 trials) and the length of time following extinction prior to test (i.e., 1 or 21 days, in a test for spontaneous recovery). We observed that outcome-specific PIT was reduced by extinction in all of our training conditions and that this extinction effect was durable, surviving a 3-week spontaneous recovery interval even though conditioned magazine approach spontaneously recovered over this interval. Although extinction reduced the magnitude of PIT, the temporal expression of PIT was mostly unaffected. We found these effects in both male and female rats, though in 1 study females were extinction-resistant. These data suggest that under the conditions studied here Pavlovian extinction may permanently weaken the ability of Pavlovian cues to retrieve a representation of their associated outcomes without impacting the temporal organization of responding. This suggests that different features of learning may be differentially sensitive to extinction.