Three expert jugglers and three intermediate jugglers performed a three-ball cascade pattern under spatially and temporally constrained conditions. In the spatially constrained conditions the balls were thrown to a specific height, whereas in the temporally constrained conditions the balls were thrown to the beeps of a metronome. The experiment was conducted to examine the hypothesis that juggling represents a spatial clock in that jugglers attempt to set up an invariant time base for the hand movements by throwing the balls consistently to a fixed height. Specifically, two expectations following from this hypothesis were examined: (1) that the spatiotemporal variability of the produced patterns would be less when juggling to an externally specified height than to an externally specified beat, because throwing to a height would be more in line with what jugglers actually do, and (2) that in both conditions the space-time trajectories of the balls would be less variable than the space-time trajectories of the hands. Examination of the observed patterns in terms of a set of theoretically motivated variables confirmed the second expectation. At the level of the individual variables the first expectation was not confirmed by the data: the spatiotemporal variability of the patterns was very similar under the two conditions. However, at the level of ensemble variables, the variability of the ball loop time (defined as the time that the ball was carried by the hand plus the subsequent flight time) was smaller when juggling to a height than when juggling to a beat, while the variability of the hand loop time (defined as the time that the hand carried a ball plus the time that it moved empty) was the same. These results were largely independent of skill level; only a few differences between expert and intermediate jugglers were found. The implications of the findings with regard to the development of a theory of perceptual-motor control in which spatial and temporal variables are linked in a task-specific manner are discussed.