1. Short term potentiation (STP) of breathing refers to respirator activity at a higher level than expected just from the dynamics of the peripheral and central chemoreceptors. In humans STP is activated by hypoxic stimulation. 2. To investigate the effects of the duration of hypoxia and the posthypoxic inspired O-2 concentration on STP, the ventilatory responses to 30 s and 1, 3 and 5 min of hypoxia (end-tidal P-O2, P-ET,P-O2 similar to 6 . 5 kPa) followed by normoxia (P-ET,P-O2 similar to 14 . 5 kPa) and hyperoxia (P-ET,P-O2 similar to 70 kPa) svere studied in ten healthy subjects. End-tidal P-CO2 (P-ET,P-CO2) was clamped during hypoxic and recovery periods at 5 . 7 kPa. 3. Steady-state ventilation (V-E) was 13 . 7 +/- 0 . 61 min(-1) during normoxia and increased to 15 . 5 +/- 0 . 31 min(-1) during hyperoxia (P < 0 . 05) due to the reduced Haldane effect and some decrease in cerebral blood flow (CBF). 4. The mean responses following hypoxia reached normoxic baseline after 69, 54, 12 and 12 s when 30 s and 1, 3 and 5 min of hypoxia, respectively, were followed by normoxia. An undershoot of 10 and 20% below hyperoxic baseline was observed when 3 and 5 min of hypoxia, respectively, were followed by hyperoxia. Hyperoxic V-E reached hyperoxic baseline after 9, 15, 12 and 9 s at the termination of 30 s and 1, 3 and 5 min of hypoxia, respectively. 5. Normoxic recovery from 30 s and 1 min of hypoxia displayed a fast and subsequent slow decrease towards normoxic baseline. The fast component was attributed to the loss of the hypoxic drive at the site of the peripheral chemoreceptors, and the slow component to the decay of the STP that had been activated centrally by the stimulus. A slow decrease at the termination of 30 s and 1 min of hypoxia by hyperoxia was not observed since this component was cancelled by the increase in ventilatory output due to the reduced Haldane effect and some decrease of CBF. 6. Decay of the STP was not apparent in the normoxic recovery from 3 and 5 min of hypoxia as a slow component since it cancelled against the slow ventilatory increase related to the increase of brain tissue P-CO2 due to the reduction of CBF at the relief of hypoxia. The undershoot observed when hyperoxia followed 3 and 5 min of hypoxia reflects the stimulatory effects of hyperoxia on V-E. 7. The manifestation of the STP as a slow ventilatory decrease depends on the duration of hypoxia and the subsequent inspired oxygen concentration. We argue that STP is not abolished by the central depressive effects of hypoxia, although the manifestation of the STP may be overridden or counteracted by other mechanisms.