1. Simultaneous measurements of mouth pressure at the end of the first 0·1 s of inspiratory occlusion ( P 0·1 ) at functional residual capacity and the maximum rate of rise of this pressure (d P /d t max.) were made repeatedly in five normal subjects during resting respiration; the coefficient of variation of d P /d t max. was 36·2%, compared with 50·6% for P 0·1 . 2. During both isocapnic hypoxia and hyperoxic hypercapnia in five normal subjects there was a close correlation between ventilation ( V̇ E ) and both P 0·1 and d P /d t max., and between end-tidal P co 2 or P o 2 and P 0·1 and d P /d t max.; during both procedures there was a close correlation between P 0·1 and d P /d t max. 3. The time at which d P /d t max. occurred ( T max. ) was not correlated with changes of d P /d t max. in either procedure. T max . was greater than 0·12 s in most studies. 4. The regression coefficients of P 0·1 and d P /d t max. on V̇ E were significantly different in hypoxia as compared with hypercapnia in four out of the five subjects; on repeated hypercapnic stimulation in two out of three subjects these regression coefficients again varied significantly. Changes in lung volume or inspiratory volume-timing relationship were not responsible for these differences. These results suggest that mouth occlusion pressure, as reflected by P 0·1 or d P /d t max., is a complex variable, reflecting the motor output of the respiratory centre, but also affected by random variations in the measurements and probably by changes in lung volume.