The aim of the present study was to examine if sleep amount prior to sleep restriction mediated subsequent task acquisition on serial addition/subtraction and reaction time (RT) sub-tasks of the Automated Neuropsychological Assessment Metric. Eleven males and 13 females [mean (SD) age = 25 (6.5) years] were assigned to either an Extended [10 h time in bed (TIB)] (n = 12) or Habitual [Mean (SD) = 7.09 (0.7)] (n = 12) sleep group for 1 week followed by one baseline night, seven sleep restriction nights (3 h TIB) and five recovery nights (8 h TIB). Throughout baseline, restriction and recovery, mathematical and serial RT tasks were administered hourly each day (08:00-18:00 h). Math and serial RT throughput for each task (speed x accuracy product) was analysed using a mixed-model anova with fixed effects for sleep group, day and time-of-day followed by post hoc t-tests (Bonferroni correction). Math throughput improved for both groups during sleep restriction, but more so compared with baseline for the prior sleep Extended group versus the Habitual group during recovery. In sum, 1 week of sleep extension improved resilience during subsequent sleep restriction and facilitated task acquisition during recovery, demonstrating that nightly sleep duration exerts long-term (days, weeks) effects.