In this paper, we introduce a multiple functional linear model with autoregressive errors. Functional linear regression, as an extension of linear regression in functional data analysis,has been studied by many researchers and applied in various fields. In many cases, data is collected sequentially over time, for example the financial series, so it is necessary to consider the autocorrelated structure of errors in functional regression background. We expand the functional coefficients on the empirical eigenfunctions of covariance operators. The expansion order increases with sample size. Under certain regular conditions, generalized least square (LS) procedure consistently estimates the functional coefficients and autoregressive coefficients. We also establish the asymptotic normality of the estimator for error's variance. A Monte Carlo simulation and real data analysis on China's weather is studied to show the finite sample performance of our proposed estimators.