In multiple logistic regression, if the explanatory variables are dependent, then we have unstable model and the estimated parameters are inaccurate. There are some methods in this thesis to overcome this problem. One of these methods, using a class of principal component estimators for logistic regression. The other one is generalized partial least squares method for logistic regression. Another method for decreasing effects of multicollinearity in logistic regression is Ridge and Stein estimation methods.