In this paper, the empirical likelihood estimation of parameter and varying-coefficient function in a varying-coefficient partial nonlinear model with measurement errors in nonparametric part and missing response data. First, we apply empirical likelihood method to construct the bias-corrected empirical log-likelihood ratio statistic of parametric and nonparametric parts based on the inverse probability weighted approach. It is proved that the constructed statistics have asymptotic chi-square distribution under some mild conditions. Based on the obtained results, the empirical likelihood confidence regions of parametric and nonparametric parts can be constructed directly. Simulation results show that our proposed method outperforms the traditional normal approximation method. Finally, we give a real data analysis.