objective: there are many ways of assessing sleepiness, which has many dimensions. in patients presenting a borderline apnea-hypopnea index (ahi, expressed as events/hour of sleep), the mechanisms of excessive daytime sleepiness (eds) remain only partially understood. in the initial stages of sleep-disordered breathing, the ahi might be related to as-yet-unexplored eds dimensions. methods: we reviewed the polysomnography results of 331 patients (52% males). the mean age was 40 ± 13 years, and the mean ahi was 4 ± 2 (range, 0-9). we assessed ten potential dimensions of sleepiness based on polysomnography results and medical histories. results: the ahi in non-rapid eye movement (nrem) stage 1 sleep (ahi-n1), in nrem stage 2 sleep (ahi-n2), and in rem sleep (ahi-rem) were, respectively, 6 ± 7, 3 ± 3 and 10 ± 4. the ahi-n2 correlated significantly with the greatest number of eds dimensions (5/10), including the epworth sleepiness scale score (r = 0.216, p < 0.001). factor analysis, using cronbach's alpha, reduced the variables to three relevant factors: questionnaire (α = 0.7); polysomnography (α = 0.68); and complaints (α = 0.55). we used these factors as dependent variables in a stepwise multiple regression analysis, adjusting for age, gender, and body mass index. the ahi-n1 correlated significantly with polysomnography (β = -0.173, p = 0.003), and the ahi-n2 correlated significantly with complaints (β = -0.152, p = 0.017). the ahi-rem did not correlate with any factor. conclusions: our results underscore the multidimensionality of eds in mild sleep apnea.