Cobb-Douglas function with additive errors is a function which can be used to analyse the relationship between production output and production factors. The method commonly used to estimate the parameter of that function is Nonlinear Least Square (NLS) and a common algorithm for this method is Gauss Newton iteration (NLS-GN). However, NLS-GN method has less-optimum results when analysing multicolinearity data. A possibly better method for this analysis is Genetic Algorithm (NLS-GA). The purpose of this study is to analyse the use of Genetic Algorithm to estimate parameters of Cobb-Douglas function with additive errors. The results show that NLS-GA method could not produce a better parameter estimator than NLS-GN method does but it produced a better parameter estimator in analysing multicolinearity data. NLS-GA method is capable of producing a better model with predictive ability than NLS-GN method does with real data. Keywords: cobb-douglas function, genetic algorithm, nonlinear least square