Abstract: The traditional training method of HMM for Web information extraction is sensitive to the initial model parameters and easy to lead to a sub-optimal model in practice.A hybrid algorithm is proposed to optimize HMM parameters by using genetic algorithm for Web information extraction.The algorithm makes use real number matrix encoding as the representation of the chromosomes，the fitness values are the results of the likelihood values，combines GA and Baum-Welch algorithm to optimize HMM parameters globally，and then to adjust the Baum-Welch algorithm parameters in GA-HMM for Web information extraction.Experimental results show that the new algorithm improves the performance in precision and recall.