This study aimed to analyze the quality of credit ratings through stock price reactions following the announcement of credit ratings after the IMF crisis. For this purpose, 297 cases of commercial paper credit rating announcements from credit rating agency A and 532 cases from credit rating agency B were selected as samples from the second half of 1998 through 2000, and the patterns of stock price reactions according to the direction and degree of rating changes and by credit rating level were examined. To summarize the analysis results: first, after the IMF crisis, the group with upgraded credit ratings showed lower stock price reactions with negative (−) signs compared to the group with unchanged credit ratings. This contrasts with the results before the IMF crisis, when the upgraded group recorded higher stock returns with positive (+) signs compared to the unchanged group, and the downgraded group recorded lower negative (−) abnormal returns. Furthermore, even when credit ratings were downgraded and showed negative (−) stock price reactions, no ordinality according to the degree of change was observed whatsoever. Finally, whereas sequential stock price reactions according to credit rating levels were observed before the IMF crisis, no such differentiation was found after the IMF crisis. These phenomena appeared identically in the credit ratings announced by the two credit rating agencies used in the analysis. These results represent phenomena where stock price reactions following credit rating announcements after the IMF crisis differ from logical inference or stock price reactions before the IMF crisis, which can be interpreted as indicating that the quality of credit ratings has relatively deteriorated. The results obtained from this study are expected to provide an impetus for enhancing the credibility of credit ratings and serve as data on the usability of credit rating information. Meanwhile, the methodology used in this study can be utilized as a model for analyzing the quality of credit ratings or exploring their usability as research data.