Nonadherence to dosing schedules for androgen deprivation therapy increases the risk of testosterone escape for patients with prostate cancer. Two approved formulations of leuprolide acetate, the most commonly prescribed androgen deprivation therapy in the United States, use different extended release delivery technologies: an in situ gel and microspheres. We evaluated the prevalence and impact of late dosing on testosterone suppression for gel and microsphere formulations of leuprolide acetate. Materials and Methods: We retrospectively analyzed records of patients with prostate cancer treated with gel or microsphere delivery of leuprolide acetate. Analyses used 2 definitions of "month," "28-day" (late dosing after day 28, 84, 112 or 168) and "extended" (late dosing after day 32, 97, 128 and 194). Frequencies of late dosing and associated testosterone values were calculated. Results: A total of 2,038 patients received gel and 8,360 received microsphere formulations of leuprolide acetate. More than 80% and 27% of injections were late for 28-day and extended month, respectively. For 28-day month late injections 10% (gel delivery) and 14% (microsphere delivery) of testosterone values were above 50 ng/dl, and 25% (gel) vs 33% (microsphere) were above 20 ng/dl. For extended month 18% (gel) vs 25% (microsphere) were above 50 ng/dl, and 34% (gel) vs 44% (microsphere) were above 20 ng/dl. Microsphere leuprolide acetate was 1.5 times more likely to have testosterone above 50/20 ng/dl vs gel. Least square mean testosterone was 34 ng/dl (gel) vs 46 ng/dl (microsphere) for 28-day month, and 48 ng/dl (gel) vs 76 ng/dl (microsphere) for extended month. Conclusions: Leuprolide acetate therapies were frequently administered late. Gel formulation demonstrated higher rates of testosterone 50 ng/dl or less and 20 ng/dl or less than microsphere formulation. Optimal testosterone suppression can impact prostate cancer progression and patient survival, and differences in extended release technology for androgen deprivation therapy appear relevant.