---

There are a lot of techniques and methods for prediction of bankruptcy among them “Statistical methods” or econometrics techniques are more popular. As dependent variable in our study is qualitative it is convenient to use qualitative discrete models. Mixed Logit model is one of the powerful and flexible techniques of discrete choices that allow the coefficients to be random with distribution function. Explanatory variables are financial ratios which derived from Zmijewski’s model. The sample data are from Tehran Stock Exchange’s Brokerage Companies during 2001-2008. We selected two random samples, one for estimation and another for prediction power test. Results show that the degree of successfulness of the model is over 90 percent.