The mass density of antecubital venous blood was measured continuously for 80 min/session with 0.1 g/l precision at a flow rate of 1.5 ml/min in six male subjects. Each person participated in two different sessions with the same protocol. To induce transvascular fluid shifts, the subjects changed from sitting to standing and from standing to supine positions. There was transient blood density shifts immediately after postural changes, followed by an asymptotic approach to a new steady-state blood density level. Additional deviations from a simple time course were regularly observed. Blood density increased by 3.5 +/- 1.4 (SD) g/l when standing after sitting and decreased by 5.0 +/- 1.2 g/l while supine after standing. The corresponding half time of the blood density increase was 5.6 +/- 1.4 min (standing after sitting) and 6.9 +/- 3.1 min (supine after standing) of the blood density decrease. Erythrocyte density was calculated and did not change with body position. Whole-body blood density was calculated from plasma density, hematocrit, and erythrocyte density, assuming an F-cell ratio of 0.91. Volume shifts were computed from the density data; the subject's blood volume density decreased by 6.2 +/- 1.2% from sitting to standing and increased by 8.5 +/- 2.1% from standing to supine. Additional discrete plasma density and hematocrit measurements gave linear relations (P less than 0.001) between all possible combinations of blood density, plasma density, and hematocrit.(ABSTRACT TRUNCATED AT 250 WORDS)