According to the fact that it is not always appropriate to assume that the model error is an independent isomorphic distribution, this paper discusses functional linear regression model with autoregressive error term. Considering the infinite dimensionality of functional variables and slope function, at first, this paper makes functional principal component analysis by taking the covariance function of functional variables as spectrum decomposition to obtain the basis function of characteristic function, then approximates the functional variables and the slope function by using this group basis respectively, so the parameters of infinite dimension are converted into basis function coefficients of finite dimension, and then constructs the parameter estimation of functional linear regression model with autoregressive error. Under certain condition, the result similar to independent isomorphic distribution is obtained. In addition, in the sense of mean square error, compared with the estimator of firstorder autoregressive error term, a smaller mean square error is received, which extends some conclusions and properties in the existing literatures.