In 29 volunteers with a low amalgam load, the number of amalgam-covered tooth surfaces and the occlusal area of the fillings were determined. Before and at select times after removal of all amalgams, concentrations of total mercury were measured by cold-vapor atomic absorption in plasma and erythrocytes as well as in urine together with the excretion rate. Absorbed daily doses were estimated from intraoral Hg emission by two separate methods. The transfer of Hg from the fillings via the oral cavity and blood to urinary excretion was evaluated according to the most representative combination of parameters. This consisted of occlusal area (1), absorbed dose (2), Hg concentration in plasma (3) and urinary excretion (4). Pairwise correlation coefficients were 0.49 for parameters 1 vs. 2, and 0.75 each for parameters 2 vs. 3 and 3 vs. 4. Within 9 days after removal of the fillings, a transient increase in Hg levels was observed in plasma only; in the group without a rubber dam, concentrations increased significantly above pre-removal values at days 1 and 3, whereas they decreased significantly below pre-removal values at day 30 in the rubber-dam group and at day 100 in both groups. Excretion rates decreased significantly at day 100 in the protected group. Peak plasma-Hg was 0.6 ng/ml on average at day 1 and decreased with halftimes of 3 and 43 days in subjects protected by a rubber dam. The results indicated that concentrations of total mercury in plasma responded rapidly to changes in the amalgam status and reflected the actual absorption most reliably. Notably, plasma-Hg levels were sensitive enough to detect a transient attenuation of the additional exposure after using a rubber dam during the removal of only a few fillings. However, being small in magnitude and lasting 100 days at best, the rubber-dam effect had minor toxicological relevance.