The output of a solar photovoltaic (PV) system can be predicted using a parameterized model. The parameters can be obtained from the datasheet of the panel or from the measured current-voltage (I-V) characteristics. Measurement-based parameter estimation is shown to provide better results, as compared with the datasheet-based method. This effect magnifies for older panels due to the deterioration of panel characteristics with time, leading to a deviation from the datasheet specifications. Leastsquares minimization is applied on the measured data using the Levenberg-Marquardt algorithm (LMA), trust region reflective Newton, and steepest descent optimization, which are compared on the basis of accuracy and speed. Low computational complexity and high parameter accuracy is ensured by introducing a novel sequential parameter estimation, in which two of the parameters are analytically evaluated, and a separate sequential evaluation of the shunt resistance parameter is performed. The predicted maximum power from the proposed method matches the experimental measurements with high accuracy.Index Terms-Gradient methods, least-squares minimization, maximum power point (MPP), parameter estimation, photovoltaic (PV) cells.