<p>In numerous application areas of regression analysis, prior information about the relationship between explanatory variables and response variables is usually available, and can be expressed as precise restrictions on regression coefficients. In this paper, under exact linear restrictions, we consider the partially linear varying coefficient model with missing response variables and measurement errors potentially in all the covariates. By combining local bias-corrected profile least squares method and Lagrange multiplier method, two types of restricted estimators of the parametric vectors and nonparametric functions are introduced based on complete-case data and an imputation technique, respectively, and their asymptotic distributions are also derived under some mild conditions. Simulation experiments are carried out to evaluate the finite sample performance of the proposed methodologies. Furthermore, a real example in epidemiology is provided for illustration.</p>