OBJECTIVES: Numerous studies have compared risk adjustment measures (RAMs), yet none have done so across various outcomes in multiple acute and chronic conditions in a single database with uniform programmatic operationalization. This study compares the performance of 7 RAMs and highlights practical programming considerations for hands-on data analysts operationalizing RAMs. METHODS: Data were administrative claims from the 2006 – 2008 MarketScan® Commercial Database. Seven RAMs (2 Deyo-Charlson Comorbidity Index variations, Chronic Disease Score [CDS], 2  3-digit ICD-9-CM code count variations, number of unique National Drug Classification [NDC] codes, and number of unique drug molecules) measured over a 1-year baseline period were compared in 7 conditions (acute coronary syndrome, sample N = 14,951; rheumatoid arthritis [RA], N = 27,085; depression, N = 129,206; diabetes, N = 126,087; hypertension, N =225,080; asthma, N = 56,172; fibromyalgia, N = 52,365) on the basis of 3 outcomes (total healthcare cost, emergency room [ER] visits, inpatient admissions) measured over a 1-year follow-up period. Goodness-of-fit statistics (chi-squared statistic for total healthcare costs and c-statistic for ER visits and inpatient admissions) were compared across age and sex-adjusted regression models for each individual RAM. RESULTS: A unique 3-digit ICD-9-CM code count that excluded ‘rule-out’ diagnoses consistently outperformed, i.e., had highest chi-squared and c-statistics, every other RAM in every condition and outcome with the exception of the number of unique NDC codes, which performed the best for all outcomes in depression and fibromyalgia patients. The number of unique NDC codes was generally the second-best performing RAM across all outcomes. The CDS performed worst across every condition and outcome. CONCLUSIONS: Complex RAMs are subject to inconsistencies in their operationalization and application from a programming perspective. Across multiple acute and chronic conditions, the two simplest and programmatically-transparent RAMs were the most predictive measures of total healthcare cost, ER visits, and inpatient admissions.