We hypothesized that: (a) the presence of microsleep (MS) during a Maintenance Wakefulness Test (MWT) trial may represent a reliable marker of sleepiness in obstructive sleep apnea (OSA) patients; (b) the number of MSs will be higher in sleepy versus non-sleepy patients with a borderline MWT mean sleep latency; and (c) scoring MS during MWT analysis may help physicians to recognize patients with a higher degree of sleepiness. We analysed the MWT data of 112 treatment-naive OSA patients: 20 with short sleep latency (SL, sleep latency <12.8 min), 43 with borderline latency (BL, sleep latency between 12.8 and 32.6 min) and 49 with normal latency (NL, sleep latency >32.6 min). Microsleep was identified in all SL, in 42 BL and in 18 NL patients, with a median latency of 5.6 min. Accordingly, patients were classified into two subgroups: group A (n = 43) with microsleep latency <5.6 min and group B (n = 69) with microsleep latency >5.6 min when present. The mean sleep latency in the MWT was 14.5 +/- 7.5 min in group A and 34.6 +/- 7.4 min in group B (p < 0.0001). The number of microsleep episodes during each MWT trial was higher in group A than in group B. Sleep latency survival curves demonstrated different patterns of sleep latency in these groups (log-rank test <0.0001). This finding was confirmed in a Cox proportional hazard analysis: the presence of a mean MS latency MWT (RR, 1.93; 95 CI 1.04-3.6; p = 0.03). We conclude that the detection of microsleep may help in discriminating OSA patients with and without daytime vigilance impairment.