The aim of this study was to develop an arterial occlusion method and compare it with the venous occlusion method for measurement of peripheral oxygen utilisation in neonates using near infrared spectroscopy (NIRS). Twenty healthy neonates were studied. Arterial occlusion was produced by inflating a neonatal blood pressure cuff to 100 mmHg for 30-40 s and oxygen utilisation (VO(2)) was calculated using the HbO(2) decrement slope following occlusion. Venous occlusion was produced by inflating the cuff to 30 mmHg for 15-20 s and VO(2) was calculated by: VO(2)=HbTx4x(SaO(2)-SvO(2)), where SaO(2) is the arterial oxygen saturation measured by pulse oximetry and SvO(2) is the venous oxygen saturation measured by NIRS. Each baby had a minimum of three arterial and three venous occlusions. Criteria were developed for acceptance/rejection of an occlusion. Using the arterial method, the mean VO(2) was 1.12 mM cm(-1) O(2)/min (S.D.=0.25), (95% CI=1.00-1.24 mM cm(-1) O(2)/min). The coefficient of variation was 6.6+/-4.1%. Using the venous method, the mean VO(2) was 1.60 mM cm(-1) O(2)/min (S.D.=0.48), (95% CI=1. 38-1.82 mM cm(-1) O(2)/min). The coefficient of variation was 12. 6+/-5.7%. The correlation between the two methods was weak (r=0.28 and r(2) was 0.08). The mean difference between the two methods was 0. 47 mM cm(-1) O(2)/min (S.D.=0.51). The limits of agreement were -0. 53 to 1.47 mM cm(-1) O(2)/min. The arterial method gives more consistent results.