Our purpose was to determine the feasibility of using salicylate (SA) concentrations in saliva as an indirect measure of SA concentrations in serum of children with juvenile rheumatoid arthritis. Sixty-four paired blood and saliva samples were obtained from 17 children, 4 to 17 yr of age (average, 9 yr), who were receiving aspirin every 8 hr. Serum total (free and bound) SA concentrations ranged from 5 to 36 mg/100 ml. The fraction of free (unbound) SA in serum, determined by equilibrium dialysis against an equal volume of buffer, rose with increasing total SA concentration. SA concentrations in saliva correlated strongly with both free and total SA concentrations in serum. The ratio of SA concentration in saliva to total SA concentration in serum rose with increasing total SA concentration in serum, whereas the relationship between SA concentration in saliva and free SA concentration in serum was essentially linear. However, there was considerable intrasubject variation of the saliva SA to serum free SA concentration ratio that could not be related to fluctuations of saliva pH. Therefore another analysis of the relationship was performed, using areas under the concentration–time curve during a dosing interval (AUCτ), which were determined from 0-, 2-, 4-, and 8-hr saliva SA and serum free SA concentrations of 12 children. These patients had serum SA concentrations at or near steady state and had provided all 4 pairs of blood and saliva samples during 1 dosing interval. The time-average saliva SA and serum free SA concentrations calculated from the AUCτ values are strongly correlated (r = 0.944, p < 0.001). The mean saliva SA to serum free SA concentration ratio determined from these data is 0.637. This ratio varied appreciably between patients (coefficient of variation, 24%; range of ratio values, 0.358 to 0.892), which limits the usefulness of salivary SA determinations as a noninvasive indirect method for monitoring serum free SA concentrations in children with juvenile rheumatoid arthritis. Clinical Pharmacology and Therapeutics (1980) 27, 619–627; doi:10.1038/clpt.1980.88