Disordered nocturnal breathing with significant arterial oxygen desaturation and sleep apnoea is a feature of extreme obesity which is often difficult to manage in the short term. We have evaluated the effect of fluoxetine, a centrally acting 5-HT re-uptake inhibitor, on sleep-breathing patterns in asymptomatic extremely obese subjects. A double-blind cross-over study was used to compare fluoxetine (60 mg for three days) to placebo. Eleven obese subjects (ten males, one female, mean weight +/- s.d. 131 +/- 2 kg) slept overnight in a sleep laboratory with the polysomnographic study recorded after an initial acclimatization night. The obese subjects had normal respiratory function and normal fully awake arterial oxygen saturation (%SaO2 97 +/- 1). Marked O2 desaturation was seen in all the subjects during sleep but the average asleep %SaO2 did not differ between the two treatment phases (placebo 90 +/- 5; fluoxetine 92 + 2%). However, fluoxetine significantly increased the minimum %SaO2 recorded during the study night either by abolishing or reducing REM sleep (placebo 73 +/- 2%; fluoxetine 81 +/- 8%; P < 0.05, 95% CI -12.3 to -2.03). Frequent hypopnoea was observed in all subjects in both REM and non-REM sleep whereas apnoea was uncommon. The total apnoea/hypopnoea index fell in six subjects during the fluoxetine night, the largest reduction being seen in the most severely affected. In five of the six the improvement was associated with the abolition of REM sleep. Total sleep time did not differ between the placebo and fluoxetine nights nor did a qualitative assessment of sleep using a visual analogue score.(ABSTRACT TRUNCATED AT 250 WORDS)