BACKGROUND: Infant swaddling is common practice in some developing countries where infant respiratory morbidity is also prevalent. Little is known about the effect of swaddling on respiratory variables in healthy infants. Such information could have important implications for respiratory diseases. AIMS: To compare respiratory rates (RR) and arterial oxygen saturations (SaO2) of healthy swaddled infants and non-swaddled infants during different conditions of sleep and arousal. SETTING: Community based, nested case control study in Ulaanbaatar, Mongolia. SUBJECTS AND METHODS: Habitually swaddled and non-swaddled infants aged 9-10 weeks taking part in a randomised controlled trial of swaddling. Respiratory rate and SaO2 were measured during quiet wakefulness, feeding, quiet and active sleep. Habitually swaddled infants were studied in swaddled and non-swaddled conditions. Habitually non-swaddled infants were studied only in the non-swaddled state. RESULTS: SaO2 was higher during awake states compared with sleep states in all groups of infants. Habitually swaddled infants had lower mean SaO2 in the swaddled compared with non-swaddled condition (96.5% vs. 96.9%, p < 0.01) but these were not significantly different from the mean SaO2 of non-swaddled infants (96.9%, minimum p = 0.22). Habitually swaddled infants in the swaddled and non-swaddled states had similar respiratory rates, but these were, in both cases, significantly lower than in habitually non-swaddled infants. CONCLUSION: Swaddling has little or no clinical effect on SaO2 or respiratory rates in healthy 9-10-week-old infants in Mongolia.