The aim of this study was to establish morning and evening reference ranges for cortisol in saliva. Another objective was to compare the concentrations of the mainly free cortisol in saliva to those of total cortisol in serum as determined with a commercial radioimmunoassay. The concentrations were determined in matched samples of saliva and serum collected at 8am and 10pm from 197 healthy volunteers. The saliva samples were stable for at least 7 days at room temperature and for 9 months at —20 °C. Reference ranges, the central 95%, were estimated to 3.5—27.0 nmol/1 at 8 am and < 6.0 nmol/1 at 10 pm. The intra-assay coefficient of variation (CV) was below 5% and total CV below 10%. The relation between the cortisol concentrations in serum and saliva was nonlinear with r = 0.86 for serum concentrations < 450 nmol/1 and r = 0.44 for serum concentrations ^ 450 nmol/1. In conclusion, the satisfactory precision of the analysis and the simple non-invasive sampling procedure suggest that saliva may be used for cortisol measurements in situations where blood sampling is difficult to perform.