OBJECTIVE The purposes of the study were to: (1) describe the aggregate strength of the relationship of arterial oxygen saturation as measured by pulse oximetry with the standard of arterial blood gas analysis as measured by co-oximetry, (2) examine how various factors affect this relationship, and (3) describe an aggregate estimate of the bias and precision between oxygen saturation as measured by pulse oximetry and the standard in vitro measures. METHODS A meta-analysis was conducted. METHODS Seventy-four studies from 1976 to 1994 met the inclusion criteria of: (1) adult study population, (2) quantitative analysis of empirical data, and (3) bivariate correlations or bias and precision estimates between pulse oximeter and co-oximeter values. RESULTS There were a total of 169 oximeter trials on 41 oximeter models from 25 different manufacturers. Studies were conducted in various settings with a variety of subjects, with most being healthy adult volunteers. The weighted mean r, based on 39 studies (62 oximeter trials) for which the r statistic and number of data points were available, was 0.895 (var [r] = 0.014). Based on 23 studies (82 oximeter trials) for which bias and precision estimates and number of data points were available, the mean absolute bias and precision were 1.999 and 0.233, respectively. Several factors were found to affect the accuracy of pulse oximetry. CONCLUSIONS Pulse oximeters were found to be accurate within 2% (+/- 1 SD) or 5% (+/- 2 SD) of in vitro oximetry in the range of 70% to 100% Sao2. In comparing ear and finger probes, readings from finger probes were more accurate. Pulse oximeters may fail to record accurately the true Sao2 during severe or rapid desaturation, hypotension, hypothermia, dyshemoglobinemia, and low perfusion states.