Changes in cerebral venous oxyhemoglobin saturation reflect changes in the balance between cerebral oxygen delivery and cerebral oxygen consumption. Invasive monitoring of cerebral venous saturation (CSVO2) has provided useful information in the management of critically ill adults at risk of cerebral hypoxia. This study describes the development and validation of a non-invasive method of measuring CSVO2 suitable for use in sick neonates using near-infrared spectroscopy (NIRS) and partial jugular venous occlusion. This technique was validated by comparison with an invasive measurement of CSVO2, co-oximetry of jugular bulb blood obtained during cardiac catheterization. Agreement between the two methods was assessed using the method of J. M. Bland and D. G. Altman. Fifteen children were studied, aged 3 mo to 14 y (median 2 y). CSVO2 by co-oximetry ranged from 36 to 80% (median 60%). The mean difference (Co-Oximeter - NIRS) was 1.5%. Limits of agreement were -12.8 to 15.9%. Three different methods of analyzing the NIRS signal were compared. The best agreement was obtained when the changes occurring during the first 5 s of partial jugular venous occlusion were studied. Greatest accuracy was seen in those subjects with least movement artifact, and we believe this technique will be reliable in sick neonates.