This study was performed to assess serum testosterone alterations induced by paradoxical sleep deprivation (PSD) and to verify their attenuation during sleep recovery (SR) based on different durations and ages. Wistar male rats aged 12 weeks for the younger group and 20 weeks for the elder group were randomly distributed into one of the following groups: a control group (cage and platform), 3-day SD, 5-day SD, 7-day SD, 1-day SR, 3-day SR and 5-day SR groups. For PSD, the modified multiple platform method was used to specifically limit rapid eye movement (REM) sleep. Differences in the testosterone and luteinizing hormone levels between the younger group and the elder group according to duration of PSD and SR recovery were analysed. Testosterone continued to fall during the sleep deprivation period in a time-dependent manner in both the younger (P=0.001, correlation coefficient r=-0.651) and elder groups (P=0.001, correlation coefficient r=-0.840). The elder group showed a significantly lower level of testosterone compared with the younger group after PSD. Upon SR after 3 days of PSD, the testosterone level continued to rise for 5 days after sleep recovery in the younger group (P=0.013), whereas testosterone concentrations failed to recover until day 5 in the elder group. PSD caused a more detrimental effect on serum testosterone in the elder group compared to the younger group with respect to decreases in luteinizing hormone (LH) levels. The replenishment of serum testosterone level was prohibited in the elder group suggesting that the effects of SD/SR may be age-dependent. The mechanism by which SD affects serum testosterone and how age may modify the process are still unclear.