The use of cerebral tissue oxygenation index (c-TOI) in a clinical setting is limited by doubts concerning the accuracy of the measurements. Since there is no gold standard, validation is difficult. Our modest aim was to quantify the precision of c-TOI doing repeated measurements by reapplying the optode several times presuming no regional differences in cerebral oxygenation. Thirty-seven premature infants were examined with several measurements of c-TOI using the NIRO 300 oximeter. Three to eight measurements were done on each infant over a period of 15 to 25 min. One-way analysis of variance was used to estimate within- and between-infant variation. The median gestational age was 27.6 weeks (23.9 to 33.0). Mean c-TOI (n = 253) was 74.6+/-8.5%. The within-infant variation was 5.2% when resiting the optode. For comparison, the between-infants variation was 6.9%, while the spontaneous 2-sec variation was 2.9%. The precision of a single measurement of c-TOI was not good. By measuring five times instead of one on each subject, the precision of the mean can be assumed to be comparable to pulse oximetry. This may be too cumbersome for clinical use, but may reduce sample size in research.