Background and objective: p-agonistic opioids cause concentration-dependent hypoventilation and increased irregularity of breathing. The aim was to quantify opioid-induced irregularity of breathing and to investigate its time-course during and after an opioid infusion, and its ability to predict the severity of respiratory depression. Methods: Twenty-three patients breathing spontaneously via a continuous positive airway pressure (CPAP) mask received an intravenous (i.v.) infusion of alfentanil (2.3 mug kg(-1) min(-1), 14 patients) or pirinitramide (piritramide) (17.9 mugkg(-1) min(-1), nine patients) until either a cumulative dose of 70 mug kg(-1) for alfentanil or 500 mug kg(-1) for pirinitramide had been achieved or the infusion had to be stopped for safety reasons. Tidal volumes (V-T) and minute ventilation were measured with an anaesthesia workstation. For every 20 breaths, the quartile coefficient was calculated (Q(eff20)V(T)). Results: Both the decrease of minute volume and the increase of Q(eff20)V(T) during and after opioid infusion were highly significant (P < 0.001, ANOVA). Patients in which the alfentanil infusion had to be terminated prematurely had lower minute volumes (P = 0.002, t-test) and higher Q(eff20)V(T) (P = 0.034, t-test) than those who received the complete dose. Changes in the regularity of breathing measured as Q(eff20)V(T) parallel those of minute ventilation during and after opioid infusion. Conclusions: Opioids cause a more complicated disturbance of the control of respiration than a mere resetting to higher PCO2. Furthermore, Q(eff20)V(T) appears to predict the severity of opioid-induced respiratory depression.