The contractile vacuole of the freshwater protozoan Paramecium multimicronucleatum is a membrane-bound exocytotic vesicle that expels excess cytosolic water. The in vitro contractile vacuole isolated from P. multimicronucleatum along with a small amount of cytosol and confined under mineral oil showed periodic rounding and slackening at fairly regular intervals. Activity lasted for over 30 min at room temperature (24-27 degrees C). The rounding of the in vitro contractile vacuole corresponded to the increased membrane tension of the in vivo contractile vacuole that occurs immediately before fluid expulsion. Unlike the in vivo contractile vacuole, the in vitro contractile vacuole did not expel fluid, since it lacked a mechanism to form a pore. The subsequent slackening of the in vitro contractile vacuole corresponded to the fluid-filling phase of the in vivo contractile vacuole that occurs at decreased membrane tension. Fluid filling occurred in the in vitro contractile vacuole only when it was isolated together with its radial arms. In vitro membrane-bound vesicles obtained by 'bisecting' (although the two parts were not always identical in size) an in vitro contractile vacuole established their own independent rounding-slackening cycles. In vitro contractile vacuole vesicles could fuse again when the vesicles slackened. The fused vesicle then showed a rounding-slackening cycle with a period closer to that of the vesicle that exhibited the shorter cycle period. An additional rounding phase of the in vitro contractile vacuole could be induced by applying suction to a portion of its membrane with a micropipette when the contractile vacuole was in its slackened phase. This suggests that maximum tension development in the contractile vacuole membrane can be triggered when tension is increased in any part of the contractile vacuole membrane. The time from the start of an extra rounding phase to the next spontaneous rounding and for subsequent rounding-slackening cycles was nearly the same as that before the extra rounding phase. This implies that there is no master pacemaker to control the rounding-slackening cycle in the contractile vacuole membrane. Severed radial arms also became vesiculated and, like contractile vacuole membranes, these in vitro vesicles showed independent rounding-slackening cycles and vesicle-vesicle fusions. Thus, membrane derived from the radial arm seems to be identical in its tension-developing properties with the contractile vacuole membrane. ATP was found to be required for contractile vacuole rounding but inhibitors of actin or tubulin polymerization, such as cytochalasin B and Nocodazole, had no effect on the in vitro contractile vacuole's rounding-slackening cycle.