The present study deals with the optimization of the COCOMO Model parameters using Simulated Annealing algorithm so as to minimize the variance in software development effort. For this data is taken from the NASA projects. The data set that has been used consists of two independent variables, viz. Lines of Code (LOC) and Effort Adjustment Factor (EAF) and dependent variable as Development Effort (DE). The results so obtained have been compared with the earlier work done by the author on Artificial Neural Network (ANN) and Adaptive Neuro Fuzzy Inference System (ANFIS), along with COCOMO results. The developed SA based model was able to provide better estimation capabilities than COCOMO, ANN and ANFIS.