A commercial radioimmunoassay (RA) for salivary cortisol was evaluated using certified reference material in water and spiked to pooled saliva in the range 2.1-89.1 nmol/L. A variance component model for describing the effects of age, body mass index (BMI), diurnal variation, gender, days of sick leave during the past year, and smoking habits was established. Reference intervals for salivary cortisol in 120 healthy individuals performing their routine work were established according to the International Union of Pure and Applied Chemistry (IUPAC) and the International Federation of Clinical Chemistry (IFCC). The method evaluation of the certified reference material in water did not show any bias of the method, i.e. recovery was 97% [CI: 94%; 100.9%]. LOD (detection limit) was 1.59 nmol/L. The ratio between analytical and within-subject variation (CVa/CVi) was 0.14, indicating that the method was adequate for measurement in healthy subjects. Reference intervals were estimated to be from 3.6 to 35.1 nmol/L for samples at the time of awakening (05.27-07.27), 7.6-39.4 nmol/L for peak level in saliva samples collected 20 min after awakening (05.47-07.47), and LOD 10.3 nmol/L for late afternoon samples (17.00-19.00). Reactivity (increase from awakening to 20 min after awakening) was estimated to be 82% [CI: -179; 345%] and recovery (decrease from 20 min after awakening to 18.00) to be 80% [CI: 51; 109%]. Eighteen percent of the subjects showed a decrease in cortisol in saliva from awakening to 20 min after awakening. Salivary cortisol was not affected by age, body mass index, gender, smoking habits or days of sick leave during the past year.