Previous research has shown that the preparation of a spatially targeted movement performed at maximal speed is different from that of a temporally constrained movement (Gottlieb et al. 1989b). In the current study, we directly examined preparation differences in temporally vs. spatially defined movements through the use of a startling stimulus and manipulation of the task goals. Participants performed arm extension movements to one of three spatial targets (20°, 40°, 60°) and an arm extension movement of 20°at three movement speeds (slow, moderate, fast). All movements were performed in a blocked, simple reaction time paradigm, with trials involving a startling stimulus (124 dB) interspersed randomly with control trials. As predicted, spatial movements were modulated by agonist duration and timed movements were modulated by agonist rise time. The startling stimulus triggered all movements at short latencies with a compression of the kinematic and electromyogram (EMG) profile such that they were performed faster than control trials. However, temporally constrained movements showed a differential effect of movement compression on startle trials such that the slowest movement showed the greatest temporal compression. The startling stimulus also decreased the relative timing between EMG bursts more for the 20°movement when it was defined by a temporal rather than spatial goal, which we attributed to the disruption of an internal timekeeper for the timed movements. These results confirm that temporally defined movements were prepared in a different manner from spatially defined movements and provide new information pertaining to these preparation differences.