OBJECTIVE: To determine the effect of adding the pharmacy cost data option to the Adjusted Clinical Groups Predictive Model (ACG-PM) when estimating future total health care costs. METHODS: Longitudinal analysis using medical and pharmacy claims data from a large state employer over a 2-year period (baseline May 1, 2001 - April 30, 2002; follow-up May 1, 2002 - April 30, 2003). Continuously eligible subjects < 65 years old at the end of the study period were selected. The total cost Predictive Resource Index from the Johns Hopkins ACG System Version 6.0 was used to predict inflation-adjusted follow-up year total costs per member (medical plus pharmacy) using baseline demographic and diagnosis information, with and without including total pharmacy cost data. Results were compared to actual follow-up year total costs by grouping actual and predicted costs ($0; $1 - $1,000; $1,001 - $5,000; $5,001 - $10,000; > $10,000) and comparing the positive predictive value (PPV) within each cost grouping. Sensitivity and specificity were also individually examined. Analyses were additionally conducted within disease-specific subgroups, including diabetes, depression, asthma, and cardiovascular disease. RESULTS: In the baseline year, approximately 70% and 75% of the 344,834 included subjects used medical and pharmacy services, respectively. Baseline total cost averaged $2,665 (median: $621) and pharmacy cost averaged $640 (median: $167). Follow-up mean actual total cost was $3193 (median: $748) and mean ACG-PM predicted costs were $2789 from both models without and with pharmacy costs (respective medians: $1638; $1635). Including pharmacy costs in the model increased the PPV, especially at high-cost groups: 40.77% to 48.74% (+ 7.97%) at > $10,000 and 23.97% to 28.18% (+ 4.21%) at $5,001 - $10,000. PPVs were higher within disease-specific subgroups and increased with inclusion of pharmacy costs, with the highest PPVs in the depression cohort (>10,000: 51.31% (without pharmacy costs) to 58.92% (with pharmacy costs); $5,000-$10,000: 29.92% to 35.31%). CONCLUSIONS: Addition of pharmacy cost data to the ACG-PM results in more accurate identification of future total health care costs, especially among high-cost members.