Three one-week controlled-release methadone formulations: polylactic acid microspheres (F-PLA) and poly(lactide-co-glycolide) microspheres (F-PLGA) with 24 and 30% methadone content, respectively, and an implant of 50:50 poly(lactide-co-glycolide): methadone, were evaluated in vitro and in vivo. The implant released the total amount of methadone in vitro while microsphere formulations released the methadone incompletely, 63% from F-PLA and 85% from F-PLGA in a week. Methadone release in vivo was estimated by deconvolution, F-PLGA giving a bioavailability >99% (methadone was totally released in 48h), while the estimated bioavailability of F-PLA was lower than expected. The bioavailability of the implant by deconvolution was around 60%, but absence of methadone in the implant indicated its complete release. These differences are due to an increase in methadone clearance after 72 h of the in vivo experimental period had passed, disturbing a good in vivo-in vitro correlation. A linear correlation between in vitro methadone release and in vivo release calculated from the amount of drug remaining within the implant, was found until the drug was completely released.