Exposure thresholds for health effects associated with elemental mercury (Hg degrees) exposure were examined by comparing behavioral test scores of 19 exposed (mean urinary Hg = 36 mu g/l) with those of 20 unexposed dentists. Thirty-six mu g Hg/l is 7 times greater than the 5 mu g Hg/l mean level measured in a national sample of dentists. To improve the distinction between recent and cumulative effects, the study also evaluated porphyrin concentrations in urine, which are correlated with renal Hg content (a measure of cumulative body burden). Subjects provided an on-site spot urine sample, were administered a l-h assessment consisting of a consent form, the Profile of Mood Scales, a symptom and medical questionnaire, and 6 behavioral tests: digit-span, symbol-digit substitution, simple reaction time, the ability to switch between tasks, vocabulary, and the One Hole Test. Multivariate regression techniques were used to evaluate dose-effects controlling for the effects of age, race, gender and alcohol consumption. A dose-effect was considered statistically significant below a p value of 0.05. Significant urinary Hg dose-effects were found for poor mental concentration, emotional lability, somatosensory irritation, and mood scores. Individual tests evaluating cognitive and motor function changed in the expected directions but were not significantly associated with urinary Hg. However, the pooled sum of rank scores for combinations of tests within domains were significantly associated with urinary Hg, providing evidence of subtle preclinical changes in behavior associated with Hg exposure. Coproporphyrin, one of three urinary porphyrins altered by mercury exposure, was significantly associated with deficits in digit span and simple reaction time. The prophyrin pooled sums of rank scores were as sensitive as the urinary Hg analyses within the cognitive and motor domains but were less sensitive for the: overall battery of tests. The reported effects were detected among dentists with a mean urinary Hg level of 36 mu g/l, which lies between the proposed biologic thresholds of 25 and 50 mu g Hg/creatinine, suggesting the need for a more comprehensive study to determine the threshold of adverse biologic effects.