OBJECTIVES: To predict healthcare costs in asthma using the EQ-5D health-related quality-of-life (HRQoL) index score. METHODS: We extracted data from the 2000-03 Medical Expenditure Panel Survey (MEPS) on adult respondents who had EQ-5D index scores. Asthma patients were identified using ICD-9 (=493) and self-report on disease (n=3,783). To account for non-random selection of positive spending, the Heckman two-step selection model was used. The EQ-5D index score was categorized into categories with 0.1 gradations.  Marginal effects (ME) were estimated to quantify the marginal changes in costs corresponding to, for instance, 0.1 increments in the EQ-5D index score, or the presence of asthma or asthma treatment. All costs were adjusted to 2003 dollars. RESULTS: The EQ-5D index score was a significant predictor of healthcare costs in all models, as were asthma, asthma treatment and other covariates (p<0.01). The Heckman model suggested a significant positive spending bias (p<0.01). For the 3 groups considered--non-asthma, asthma and asthma with treatment--the mean EQ-5D index scores were 0.848, 0.755 and 0.706, respectively.  The actual mean costs were $2,355, $4,284 and $5,577, respectively. The predicted mean costs using Heckman model were $2,673, $4,431 and $5,618, respectively.  On average, a 0.1 unit improvement in EQ-5D score was associated with $502 cost reduction in asthma and $693 cost reduction if asthma patients had treatment (p<0.01). However, after adjusting for positive spending bias, the cost reduction was $891 and $961 (p<0.01) for these two groups respectively. Greater cost reductions were associated with improving EQ-5D scores at the lower end (<0 to 0.3) and the mid-range (0.5 to 0.7). CONCLUSIONS: HRQoL is an important component of healthcare resource utilization. The findings suggest a significant relationship exists between healthcare costs and HRQoL data. The EQ-5D significantly predicts costs in asthma and asthma treatment, especially after adjusting non-random positive healthcare spending.