It has been shown that logistic regression analysis has some undesirable results when outliers exist. The design of robust analysis has been studied in the literature of statistics for over two decades. More recently various robust logistic regression models have been proposed for processing noisy data. We proposed a new method using fuzzy complement and derive improved algorithms that may produce better logistic regression analysis from the spoiled data set. Experimental results show that the proposed robust method improves the performance of traditional regression on the test data when outliers exist.