OBJECTIVE: The California Medicaid program (Medi-Cal) initiated open access to atypical antipsychotic medications in October 1997. This analysis evaluates the impact of open access on drug therapy outcomes and treatment costs in patients with bipolar disorder. METHODS: We conducted a retrospective database analysis of patients diagnosed with bipolar disorder who initiated an episode of drug therapy between July 1994 and December 31, 1999. Episodes were required to have a minimum of 6 months of pre-treatment and 12 months of post-treatment data. Episodes were then separated into three time periods: a "closed-access period" prior to October 1997, a "transition period" covering the first 6 months of open access, and an "open access period" for episodes initiated after April 1, 1998. Multivariate regression models were used to estimate the impact of open access on total health care costs and duration of therapy. COX proportional hazard models were estimated for time to discontinuation. RESULTS: The number of patients re-starting drug therapy or augmenting an existing therapy increased immediately with open access due to increased use of second-generation medications. Episode initiation rates returned to pre-open access levels within 6 months (transition period). Open access significantly reduced total costs primarily due to significant savings in nursing home care ($1700 and $1807 for re-starters and augmenters, respectively). However, drug persistence also declined with open-access: 42 fewer days for re-starters and 33 fewer days for augmenters. Augmenters and re-starters were 16% and 12% more likely to discontinue therapy if their episodes were initiated in the open access period. CONCLUSIONS: The decision to include atypical antipsychotics by the Medi-Cal program resulted in lower persistency of drug therapy and lower total health care costs. Decision-makers and program administrators must use caution in evaluating the impact of open access to new antipsychotic medications on patient outcomes and costs.