1. Correlations between latency and percent correct measures on the T-maze. All significant sleep versus performance comparisons described in the text are shown for the T-maze (lines 2–6) and spatial mazes (lines 7–9). To correct for multiple comparisons an α = 0.01 was selected. No waking parameters were significantly correlated with T-maze or eight-box maze performance metrics, although there was a positive trend between quiet waking and procedural memory (Figure 2d). 2. The amount of REM sleep during the baseline recording was most correlated to the percent of correct choices and speed of trial completion on the T-maze. 3. The amount of quiet sleep (QS) during desipramine administration was correlated with both the percent correct choice and speed of completing the T-maze trial. 4. The larger the increase in QS after desipramine, the faster the T-maze trial completions. 5. The more QS obtained under DMI, the faster the trial completions. 6. The more REM sleep at baseline, the faster the trial completions. 7. On the eight-box maze the degree of improvement in the novel maze under desipramine treatment was most correlated with the preservation of the transition to REM sleep state as shown in Figure 2e. 8. Those animals whose performance on the familiar maze was best preserved under desipramine treatment were those with the smallest amount of TR sleep, as shown in Figure 2c. 9. Those animals with the largest desipramine suppression of REM sleep had the largest losses during familiar maze reconsolidation (Figure 2b). 10. Improvement on the Novel maze was not significantly correlated with any other sleep parameter, e.g. REM sleep, as listed.