In the present study an attempt was made to determine the relative effectiveness of various warm-up activities in eliminating postrest warm-up decrement (WUD) in the tennis serve. Seventy highly-skilled players hit 20 serves, rested for either 5 or 15 min, and then attempted 4 final serves. During the last 2 min of the rest period, players continued to rest, ran in place, engaged in mental imagery, performed practice swings, or repeatedly hit the ball against the ground and caught it. In addition to estimates of serving accuracy, measures of somatic and cognitive arousal were obtained at the beginning and end of the rest interval. Multiple regression procedures revealed that reductions in WUD were significantly related to the restoration of prerest arousal levels. Between-group comparisons indicated that practice swings were the most effective warm-up activity for restoring somatic and cognitive arousal to prerest levels and for eliminating WUD. Theoretical discussion centered on possible applications of Nacson and Schmidt's (1971) activity-set hypothesis to the tennis serve.