1. It is widely believed that the lung is an important site of granulocyte margination and releases most of the granulocytes of the peripheral neutrophilia of exercise. 2. We measured granulocyte margination in the lung in terms of the lung total blood granulocyte pool and the lung circulating granulocyte pool in eight patients without inflammatory disease or evidence of lung pathology by comparing the regional gamma-camera lung count rate of In-111-labelled granulocytes with that of In-111-labelled erythrocytes. According to the respective In-111 activities in peripheral blood samples taken between 5 and 40 min after granulocyte injection, the lung marginating granulocyte pool was 0.78 (SEM 0.045) of the lung total blood granulocyte pool or 4.6 (0.92) of the lung circulating granulocyte pool 5 min after injection, decreasing to plateau values of 0.57 (0.053) and 1.53 (0.28) from 20 min after injection. This compared with corresponding whole-body ratios of about 0.6 and 1.5, respectively. 3. After 4 min of maximal exercise in four normal subjects given In-111-labelled granulocytes 60 min before exercise, the In-111-labelled granulocyte count rate over the lung increased to 1.23 (0.05) of the pre-exercise value with a time course that was essentially identical with the time course of the peripheral native neutrophilia. The spleen In-111 signal decreased with the same time course, reaching a minimum of 0.63 (0.05) of the pre-exercise level at 5-10 min after the end of exercise. 4. In a further four normal subjects given Tc-99m-labelled erythrocytes, exercise resulted in an increase in the lung Tc-99m count rate to 1.11 (0.05) of the pre-exercise value. This increase was maximal immediately after the end of exercise. Based on changes in peripheral haemoglobin, total blood volume decreased in the eight normal subjects to 0.89 (0.01) of the pre-exercise value. 5. Using (a) the mean increases seen in lung In-111 and Tc-99m count rates in the normal subjects, (b) the mean haemoconcentration recorded in the normal subjects and (c) the mean baseline ratio of lung marginating granulocyte pool/lung total blood granulocyte pool measured in the patients, we calculated that the lung marginating granulocyte pool after exercise declined to only 0.94 of the pre-exercise value in the four normal subjects given In-111-labelled granulocytes. 6. We conclude (a) that although granulocytes marginate in the lung, they do not do so to an extent greater than the average for the whole body, and (b) that there is no significant release of granulocytes from the lungs as a result of exercise.