Pulse oximetry is a powerful tool for patient monitoring but there are practical problems with its application in small children. Standard probes may be too large to be used easily and dedicated paediatric probes may be awkward to apply. The authors describe a modification of the Nellcor 'Oxiband' probe which simplifies its use in children. The modified probe was compared in 27 children with the conventional 'clothes-peg' type of device and found to under-read by 1.5% with 95% confidence intervals of +/- 3.1%. This is comparable to the error of the 'Oxiband' probe in normal use of +/- 3% over the range of saturations 70-95%. It is concluded that the modified probe may safely be used in clinical practice.