OBJECTIVES: Genotype-guided therapy has been demonstrated to enhance the accuracy of warfarin dosing. However, factors that limit the predictive accuracy of this approach have not been evaluated in a heterogeneous population.  This study identified patient characteristics and clinical factors that contributed to poor prediction of a warfarin therapeutic dose when using a genotyped-guided approach. METHODS: This was a prospective observational study of patients managed by the University of Illinois Hospital and Health Sciences System Pharmacogenetics Service who have attained a warfarin therapeutic dose. The primary outcome was a large dose deviation (>20%) between the genotype-guided predicted dose and the actual warfarin therapeutic dose.  Variables examined included patient demographics, interacting medications, comorbidities, genotype information, baseline international normalized ratio , and reason for anticoagulation. Multivariate logistic regression was used to estimate the odds ratio (OR) for a large dose deviation. RESULTS: Of the 146 consecutive patients enrolled, 23.3% were Hispanics, 54.1% were African American, and the mean age was 54.2±17.9 years.  Fifty-three percent of the study cohort had a large dose deviation. After adjusting for potential confounders, being middle aged (41-65 vs. 18-40 years) and Hispanic (vs. Whites and Asians) had a 4.15-fold (95%CI 1.24-13.95, p=0.021) and 7.58-fold (95%CI 1.69-33.93, p=0.008) increased odds for a large dose deviation, respectively.  Patients who took warfarin for thrombosis prophylaxis versus treatment were 3.2-fold more likely to have a large dose deviation (95%CI 1.14-8.99, p=0.027).  In addition, patients who required a high warfarin maintenance dose (>7.5mg/day) versus average dose (3-7.5 mg/day) were at an elevated risk (adjusted OR 5.21, 95%CI 1.30-20.78, p=0.019) of having a large dose deviation. CONCLUSIONS: Our results highlight factors that might contribute to poor prediction of warfarin dosing using a genotype-guided approach and suggest that methods to refine dose prediction are needed for under-studied patient populations, including Hispanics and patients receiving warfarin for thrombosis prophylaxis.