Experimental research has shown correlation between near infrared spectroscopy (NIRS) and blood loss, but these findings have not been validated in man. Ten blood donors were monitored before, during and for 10 min after blood collection (470 ml) with NIRS. A Somanetics INVOS 4100 oximeter monitored regional haemoglobin saturation in the cerebral cortex (cSO(2)-left frontal area) and from the left calf (pSO(2)). A Critikon 2001 Cerebral Redox Model monitored total (tHb), oxygenated (O(2)Hb) and deoxygenated (HHb) haemoglobin from the right calf. The oxygenation index [HbD]=[O(2)Hb]-[HHb] was derived from the data. cSO(2) (P<0.001), pSO(2) (P<0.001) and HbD (P=0.001) decreased during blood collection. Maximum changes occurred 10 minutes after collection for cSO(2), with a mean fall (95% C.I.) of 2.5 (-0.06-4.86)%, at the end of blood collection for pSO(2), with a mean fall (95% C.I.) of 3 (0.74-5.26)% and after 8% of blood volume loss for HbD, with a mean fall (95% C.I.) of 7.2 (2.25-12.16). Cerebral and peripheral oxygenation did not recover after blood collection. There was good correlation between NIRS parameters and blood loss. NIRS is a potentially useful technique for monitoring blood loss in humans. Further research is needed to define its role in clinical practice.