Operant extinction is learning to supress a previously rewarded behavior. It is known to be strongly associated with the specific context in which it was acquired, which limits the therapeutic use of operant extinction in behavioral treatments, e. g., of addiction. We examined whether sleep influences contextual memory of operant extinction over time, using two different recall tests (Recent and Remote). Rats were trained in an operant conditioning task (lever press) in context A, then underwent extinction training in context B, followed by a 3-h retention period that contained either spontaneous morning sleep, morning sleep deprivation, or spontaneous evening wakefulness. A recall test was performed either immediately after the 3-h experimental retention period (Recent recall) or after 48 h (Remote), in the extinction context B and in a novel context C. The two main findings were: (i) at the Recent recall test, sleep in comparison with sleep deprivation and spontaneous wakefulness enhanced extinction memory but, only in the extinction context B; (ii) at the Remote recall, extinction performance after sleep was enhanced in both contexts B and C to an extent comparable to levels at Recent recall in context B. Interestingly, extinction performance at Remote recall was also improved in the sleep deprivation groups in both contexts, with no difference to performance in the sleep group. Our results suggest that 3 h of post-learning sleep transiently facilitate the context specificity of operant extinction at a Recent recall. However, the improvement and contextual generalization of operant extinction memory observed in the long-term, i. e., after 48 h, does not require immediate post-learning sleep.