OBJECTIVES: This study aimed to estimate the withdrawal rate from the labor market among patients with schizophrenia or bipolar/depression in Taiwan. METHODS: The data source was the Psychiatric Inpatients Medical Claims Data (PIMC) from the National Health Research Institute (NHRI), Taiwan. The PIMC compiled all the health records during 1996-2008 for patients who had at least one psychiatric hospitalization during 1996-2001. The inclusion criteria were patients who: 1) had their initial psychiatric health record within 1998 to 2001; 2) had primary or secondary ICD-9-CM diagnosis of schizophrenia, bipolar or depression; 3) were under employment; and 4) were aged between 18 and 65. The final sample available for the analyses was 23,174 (11,970 for schizophrenia; 11,204 for bipolar/depression). An index date was created by subtracting 365 days from the date of initial health record for examining the impact of the disease on withdrawal pre and post of disease onset. To identify all withdrawal events, each case was tracked from the index date until December 31, 2008 or death, whichever came first. Women without events were censored on December 31, 2008. Kaplan-Meier method was used to estimate cumulated employment rates. RESULTS: For patients with schizophrenia, withdrawal rate of 34% was found during one year before the disease onset. During one year before and one after the disease onset, the withdrawal rate increased to 52%. During one year before and four year after the disease onset, the rate raised to 78%. Similar patterns were observed for patients with bipolar/depression, with the corresponding rate equaling 34%, 52%, and 75%. Median employment time (95% CI) was 672.0 days (645.0, 698.0) for patients with schizophrenia, and 689.5 days (662.0 to 712.0) for patients with bipolar/depression. CONCLUSIONS: Schizophrenia, bipolar and depression were showed to have adverse impacts on labor participation.