The present paper focuses on human studies attempting to relate sleep states to memory processes. These studies typically present learning material to participants and then examine their ability to recall this material after intervening post-training sleep or sleep deprivation. Most experiments utilize either sleep recording or sleep deprivation following task acquisition to reach their conclusions, although cueing and position emission tomography (PET) scan studies have also been done. Results strongly suggest that REM sleep is involved with the efficient memory processing of cognitive procedural material but not declarative material. Although there are some data to suggest that stage 3/4 or NREM sleep is necessary for declarative memory consolidation, NREM may in fact simply be occurring at the same time as another factor that is actually involved in the memory processing. Preliminary results suggest that the length of the NREM-REM sleep cycle may be important for declarative memory. Preliminary data also suggest that stage 2 sleep may be involved with the memory for motor procedural but not cognitive procedural tasks. Sleep researchers would do well to capitalize on the latest advancements in memory research by choosing tasks that represent special memory systems and examining their relationships to sleep states.