Publisher Summary This chapter presents the results of applying various neural network systems as well as chosen statistical methods to firm classification. Statistical data concerning 115 small enterprises that applied for a loan in two regional banks in Lodz are used in investigation. The aim of the research is to analyze the efficiency of methods in the loan granting procedure. The data sets that are used in the training process are extremely important. It is necessary to have a wide representation of all probable cases which requires a huge number of observations. The results of investigation show that neural networks can be successfully used as an alternative method in the loan decision making procedure. The smallest classification errors are obtained using multilayer perceptrons. RBF networks generate higher general classification errors but the majority of them correctly recognize uncreditworthy enterprises. It is suggested that supervised classification methods are more advisable than self-organizing networks.