OBJECTIVES: To examine sources of error in claims-based adherence calculations for LAI antipsychotics with potentially invalid days’ supply (DS) values and evaluate the assumption that quantity-dispensed (QD) values are in product units. METHODS: Pharmacy claims for single-dose LAI antipsychotics dispensed between January 1, 2009 and December 31, 2010 were selected from a large US database. Frequency distributions were generated for observed DS and QD values for each product and dose. Observed QD values on premixed LAI antipsychotic claims were divided by the product’s volume to test the assumption that QD was entered in milliliters rather than units. After adjustment to QD for premixed LAI antipsychotic claims, duration of therapy per injection was calculated for all LAI antipsychotics as DS/QD. Calculated therapy duration was compared with the dosing interval in the product’s package insert (PI). Percentage of claims with duration of therapy per injection within the product’s PI range was calculated as a measure of the validity of the observed DS value. RESULTS: For the 611,325 LAI antipsychotic claims analyzed, observed QD values ranged from 0.01 to 117, suggesting values that did not always represent product units. After adjustment to QD for premixed LAI antipsychotics, 98.5% of claims had an integer value for  calculated quantity in product units, supporting the assumption that premixed LAI antipsychotics’ quantities were entered in milliliters. After adjustment, 21.5% of claims had a calculated therapy duration per injection outside the PI range. Percentage of claims with calculated therapy durations outside the PI ranged from 10.6% to 39.1% for paliperidone palmitate, 7.6% to 13.1% for risperidone long-acting injection, and 3.1% to 10.8% for olanzapine pamoate. CONCLUSIONS: Results raise concerns regarding potentially invalid values in DS and QD fields. Algorithms for appropriate use of LAI antipsychotic pharmacy claims in adherence calculations, quality measurement, and cost analyses are recommended.