A nonparametric local linear estimator of the regression function when both the response and the explanatory variables are of the functional kind is constructed. Then its rate of uniform almost-complete convergence is stated. This theoretical result will be a key tool for many further developments in nonparametric functional data analysis (FDA).