Although long-acting injectable risperidone (LAIR) has been hypothesized to improve medication adherence compared to oral medications, data from real-world practice have yet to be presented on time to treatment discontinuation. Records of all new prescriptions for antipsychotic medication during the first 2 months of FY 2006 for VA patients diagnosed with schizophrenia (N = 11,821) were examined and duration of treatment with LAIR and oral antipsychotics were calculated for the next 2 years. Multivariable logistic regression was used to identify patient characteristics independently associated with receipt of LAIR. Proportional hazards models were used to compare the likelihood of discontinuing each of the medications as compared to LAIR. Altogether, 2.4% of the 11,821 new starts were prescribed LAIR, 44.6% of whom continued therapy for 540-720 days (18-24 months), less than the 77.1% of those on clozapine, 57.9% on oral conventional antipsychotics, 55.0% on olanzapine, and 49.5% on risperidone, but more than the 27.7% on aripiprazole. After adjusting for potentially confounding factors, patients who were initiated on LAIR were more likely to discontinue their medication than those who were initiated on oral first- or second-generation antipsychotics (SGAs) with the exception ziprasidone and aripiprazole. Less than half of patients on LAIR continued treatment for 18 months, a smaller proportion than of those started on most oral first- or second-generation antipsychotics, suggesting that for many patients with schizophrenia improved adherence from this treatment may not be sustained.