We investigate the serial correlation test for partially functional linear regression models (PFLRMs). PFLRMs relate a scalar response to predictor variables that composed of multivariate vectors and random functions. We propose the V_T,p and empirical log-likelihood ratio statistics and derive the asymptotic distribution of them under the null hypothesis. Numerical study with finite sample size reveals that the proposed statistics and procedures have good performance. Finally, we explore the application of the tests on the Berkeley growth data, results show that our test statistics are effective and stable.