This document describes a study that used a multi-state Markov model to analyze patterns of use of three opioid dependence treatments: methadone, buprenorphine, and other drug-free outpatient treatments. The study followed 30,080 Medicaid patients diagnosed with opioid dependence from 2003 to 2007. The model estimated transition probabilities between the three treatment states and discontinuation over time. Results found the probability of discontinuing methadone treatment was 20% after 6 months, while survival probabilities on each treatment after 30 months were 30% for methadone, 9% for buprenorphine, and 5% for other treatments. The study concluded multi-state Markov models are useful for investigating treatment switching and compliance over long