Abstract: In order to predict the probability of company default accurately and efficiently, we take the logit regression model with single variable model and multi-variable model to test each enterprise and inference the probability of default. By taking German company as an example, to analysis the common and specific characteristics of the default company. The results show that, measuring the statistics approach to predict company default by logit model has a high degree of accuracy of 70% , maintains high reliability and stability. Default prediction can greatly reduce the risk of default in the field of financial investment with an important role.

Key words: Logit regression, credit default, default probability