The objective of this research is to compare four discriminant models FLDF: Fisher's Linear Discriminant Function, LRA: Logistic Regression Analysis, MSD (Freed and Glover, 1986), and LCM (Lam, Choo, and Moy, 1996) for two-group discriminant analysis. The data were financial ratios of 210 Thai listed companies collected during 2536-2539 B.C., 30 bankrupts and 180 non-bankrupts. The performance of each model was assessed by correct classification and prediction accuracy. The result is the LRA model performed best, followed by the MSD and LCM models respectively. In terms of prediction accuracy, the LCM model performed best, followed by the LRA and MSD models respectively.