A novel weighted Root Mean Square Deviation (RMSD) control model based on the equally weighted sum of errors of the maximum, mean, and minimum values in each forecasting time period has been developed and applied with Holt-Winters forecasting method to do days or weekly ahead forecasting of daily averaged solar irradiance, output power, and energy efficiency for grid connected solar photovoltaic (PV) systems. The forecasting results based on the present model are compared favorably with the measured data of a grid connected solar PV system in Macau for solar irradiance, output power, and the system energy efficiency. Comparisons of the predicted results based on the present weighted RMSD control model to those from previous absolute error and relative error control models show that present weighted RMSD control model has better stability and higher accuracy.