Objectives We studied sleep in a rural population in Madagascar to (i) characterize sleep in an equatorial small‐scale agricultural population without electricity, (ii) assess whether sleep is linked to noise levels in a dense population, and (iii) examine the effects of experimentally introduced artificial light on sleep timing. Methods Using actigraphy, sleep–wake patterns were analyzed for both daytime napping and nighttime wakefulness in 21 participants for a sum total of 292 days. Functional linear modeling was used to characterize 24‐h time‐averaged circadian patterns and to investigate the effect of experimentally introduced mobile field lights on sleep timing. We also obtained the first polysomnography (PSG) recordings of sleep in a traditional population. Results In every measure of sleep duration and quality, the Malagasy population experienced shorter and lower quality sleep when compared to similarly measured postindustrial values. The population slept for a total of 6.5 h per night and napped during 89% of recorded days. We observed a peak in activity after midnight for both sexes on 49% of nights, consistent with segmented sleep. Access to mobile field lights had no statistical effect on nighttime sleep timing. From PSG, we documented relatively short rapid eye movement (14%), poor sleep efficiency (66%), and high wake after sleep onset (162 min). Conclusions Sleep in this population is segmented, similar to the “first” sleep and “second” sleep reported in the historical record. Moreover, although average sleep duration and quality were lower than documented in Western populations, circadian rhythms were more stable across days.