For the linear discrete time-invariant multisensor system with uncertain model parameters and noise variances, by using the approach of compensating the parameter uncertainties by a fictitious noise, the system is converted into a system with uncertain noise variances only. By using the weighted least squares(WLS) method and the minimax robust estimation principle, based on the worst-case conservative system with the conservative upper bounds of noise variances, a robust weighted measurement fusion Kalman predictor is presented, and its robustness is proved by using the Lyapunov equation approach. The accuracy comparisons among the robust local Kalman predictors, weighted measurement fusion Kalman predictor and centralized fusion Kalman predictor are given. A simulation example is presented to demonstrate how to search the robust region of uncertain parameters and to show the effectiveness and correctness of the proposed results.