The robust spatio-temporal signal parameter estimation is studied in this paper in the case of general errors. First a general array error model is given. Then a spatio-temporal weighted subspace fitting (ST-WSF) method is proposed on the basis of the given model. And the Cramér-Rao lower bounds on signal parameters, i.e. directions of arrival and carrier frequencies, are derived. Numerical experiemtns further verifies the effectivments of the algorithm.