The aim of this study was to examine the influence of restricted knee motion during the serve in tennis players of different performance levels. Thirty subjects distributed in 3 groups (beginner, B; intermediate, I; elite, E) performed 15 flat first serves with normal (normal serve, S(N)) and restricted (restricted serve, S(R)) knee motion. In S(R), the legs were kept outstretched by splints with a knee joint angle fixed at 10 degrees (0 degrees fully extended) to prevent any knee flexion/ extension. Vertical maximum ground reaction forces (Fz(max)), ball impact location (L(impact)), and ball speed (S(ball)) were measured with force platform, video analysis, and radar, respectively. Fz(max), L(impact,) and S(ball) were higher (p < 0.001) in S(N) than in S(R). S(ball) was significantly (p < 0.001) dependent on performance level, with higher values recorded in E than in B or I. From S(R) to S(N), increase in L(impact) was greater (p < 0.01) in E than in other groups and increases in Fz(max) and S(ball) were correlated (r = 0.69, p < 0.01) in E only. Knee motion is a significant contributor to serving effectiveness whatever the performance level. Skilled players perform faster serves than their less skilled counterparts, and this is partly related to a more forceful lower limb drive.