We intended to explore the percentages and session times of psychotherapy for completers, informed dropouts, and non-informed dropouts, and to analyze the differences of clinical variables among those groups, and in an attempt to fi nd effective predictors. Methods: The study data were based on fi ve-year process note archives of short-term individual psychotherapy at a specialty psychiatric hospital. We did analysis of variance and multi-nominal logistic regression to analyze 16 clinical variables of 97 adult patients. Results: The percentages of completers, informed dropouts, and non-informed dropouts were 34%, 36.1%, and 29.9%, respectively, of the sample population. Those percentages are not signifi - cantly different in three groups. Their average times of attended sessions were 13.12, 6.69, and 6.31 times among those three groups, showing signifi cant difference (F = 66.3, df = 2, 94, p < 0.001). The factor of sex (χ2 = 9.55, p < 0.01), age (F = 3.99, df = 2, 92, p < 0.05), and past psychotherapy experience (PPE) (χ2 = 9.60, df = 2, p < 0.01) could effectively discriminate three patients groups, with a total correct prediction rate of 55.8%. PPE was the most signifi cant predictors among three predictors, but it could not effectively discriminate between completers and informed dropouts. Conclusion: Being young, male and without PPE were risk factors for non-informed dropout. Future studies are encouraged to set up longitudinal and detailed psychotherapy archives to further explore different features of dropout subgroups.