The aims of this study were to examine the relationship between anthropometric variables, physical performance, and functional test with serve velocity regarding tennis players' level and to design regression models that effectively predict serve velocity. A sample of sixteen male tennis players participated in this study (national level = 8, professional level = 7). Anthropometric measurements (body mass, height, body mass index and body segments) and physical test (hand strength, countermovement jump, jump on serve, and serve velocity) and functional test (medicine ball throw overhead and shot put) were performed. No differences in anthropometrics and physical test were found between national and professional levels. A significant positive correlation (p < 0.05, ranging for 0.603 to 0.932) was found between some anthropometrics measurements (body mass, height, arm, forearm, and leg segments), physical parameters (hand strength, countermovement jump) and functional test (medicine ball throw shot put and overhead) with serve velocity for all tennis players. Multiple regression analysis indicated that medicine ball throw shot put was the most important test to explain serve velocity (r2 = 0.869). The results showed how the combination of physical and anthropometric factors have an impact on serve velocity. In addition, a new functional fitness test (medicine ball throw shot put) is proposed as an alternative to traditional medicine ball throw overhead due to its high reproducibility (inter-trial reliability) and predictive validity values, as well as by multi-segmental coordination movement similar to tennis serve.