The aim was to determine the relationship between mercury content of resting and stimulated saliva, and blood and urine. Eighty subjects participated; 40 of them attributed their self-reported complaints to dental amalgam (patients), the others were matched with respect to age, sex and amalgam restorations (controls). Serum, 24-h urine, resting and chewing stimulated saliva were analyzed for mercury using the ASS-technique. Quality, number, surfaces and total area of amalgam fillings were recorded clinically and using study models. Median (range) mercury levels in serum were 0.67 (0.1-1.52) microgram/l for patients and 0.60 (0.1-1.3) for controls. In urine levels were found to be 0.77 (0.11-5.16) and 0.94 (0.17-3.01) microgram/g creatinine respectively. No significant differences were found between the groups. Resting saliva contained 2.97 (0.10-45.46) micrograms/l in patients and 3.69 (0.34-55.41) in controls (not significant). Chewing mobilized an additional amount of 16.78 (-6.97 to 149.78) micrograms/l in patients and 49.49 (-1.36 to 504.63) in controls (P < or = 0.01). Only a weak correlation was found between mobilized mercury in saliva and serum (r = 0.27; P < or = 0.05) or urine (r = 0.47; P < or = 0.001). For resting saliva the respective values were r = 0.45 (P < or = 0.001) and r = 0.60 (P < or = 0.001). Saliva testing is not an appropriate measure for estimating the mercury burden derived from dental amalgam.