In this paper, a method for automatic word sense disambiguation based on logarithm model (LM) is discussed, and a word sense disambiguation system LM_WSD is implemented. In the experiments, four models are used to word sense disambiguation. Experiments showed the effect of high-frequency sense, salient words, specialized field and general usage to noun and verb word sense disambiguation. Now the system LM_WSD was applied in a word based English-Chinese machine translation system for car fittings field, and improved the performance of the system.