Background: Patient genotype influences outcome of HIV infection and treatment. In a pilot project, we test implementation (operational, semantic, procedures for delivery), doctor’s perception and predictive power of a custom genotyping array that synthesizes current knowledge on genetic variants that contribute to clinical phenotypes in HIV. Methodology: The array includes 4,146 SNPs that capture known functional variants and tags common variation (>5% frequency) in relevant genes and also allows prediction of class I HLA types. Of these, 81 SNPs are reported markers associated with HIV/HCV disease progression, adverse drug events, pharmacokinetics or metabolic traits. Results: In this first interim analysis, 89 HIV infected patients were enrolled prior to the initiation of cART. Genotyping was performed and reports of genetic predisposition were returned to clinicians after initiation of treatment. Of these, 84 (94%) carried genetic risk factors for at least one trait. To assess predictive power, we explored metabolic parameters and drug levels in detail. A total of 203 metabolite measurements were available for bilirubin, HDL cholesterol, non-HDL cholesterol, triglycerides and glucose and 28 measurements of drug levels for which genetic predictions exist (EFV, ETV and LPV/r). We observed 39 metabolite measurements that exceeded the threshold for reporting. In 5 individuals (13%), relevant genotypes were observed and correct predictions made. Conversely, 13 of 164 individuals (8%) carried relevant genetic markers without the associated trait. We observed 6 drug levels that were at or above the 75th percentile. In 3 individuals (50%), relevant genotypes were observed and correct predictions made. Conversely, 1 of 22 individuals carried relevant genetic markers without the associated trait. In a survey aimed at gauging the utility of this information as seen by the physician, 57% said they found the information useful or potentially useful. In 7 of 57 (12%) instances, the treating physician would have been inclined to prescribe a different treatment, 9 of 58 (16%) would consider additional laboratory tests and 20 of 58 (34%) would discuss cardiovascular risk factors. Conclusions: The interim analysis demonstrates the feasibility and general acceptance of returning genetic data to clinicians and the modest predictive power of the genetic testing. Based on physician responses, these tests could have lead to a change in clinical decisions.