OBJECTIVES: Design a predictive model which determines the probability of patients using a drug typically prescribed later in the therapeutic management of HIV/AIDS.  Determining when a patient may need to switch to a different antiretroviral regimen for HIV/AIDS due to complexities (e.g., resistance, poor adherence, etc.) remains challenging.  Personalized approaches for the therapeutic management of HIV/AIDS are needed. A national database of pharmacy claims was analyzed to develop a predictive model that can be used with traditional measures for regimen selection. METHODS: Beginning April 2006 to March 2009, prescription claim data for 15,828 patients with continuous benefit eligibility were analyzed for; antiretroviral drug regimen, presence of co-morbid illnesses (across 50 diseases), and basic demographics.  Regimen identification was based on drug combinations included in the 2008 DHHS Guidelines for the Use of Antiretroviral Agents in HIV-1 Infected Adults and Adolescents.  Late-type regimens included enfuvirtide, etravirine, maraviroc, or raltegravir.   A multivariate logistic regression model was utilized across drug-centric and patient-centric domains to analyze the likelihood of a patient being prescribed a late-type regimen. RESULTS: N=15,828 patients initially included in the model and n= 1,838 (11.6%) identified with a claim for the target drug(s).  Five or more drug regimens (OR=32.81; CI: 28.29- 38.04) and three or more average number of drugs per regimen (OR=4.13; CI: 3.15 – 4.87) are the strongest predictors and dominate the model.  Cancer (OR=1.80; CI: 1.31-2.32), type 2 diabetes (OR=1.74; CI: 1.38-2.20), acne (OR=1.56; CI: 1.09 – 2.22), and Benign Prostate Hyperplasia (OR=1.54; CI: 1.14-2.08) were also significant.  CONCLUSIONS: Ability to develop and apply a highly predictive model identifying patients suitable for drugs typically prescribed later in therapeutic management of HIV/AIDS exists.  Application of this model can lend significant insight into patient management and further therapeutic precision.  Additional studies are needed to confirm the clinical benefit of this model.