OBJECTIVE: To derive health status values (utilities) for the Swedish population. METHODS: EQ-5D data were collected through a mailing survey to a random sample (n=2990) of the Swedish population (response rate 65%: complete data on 1741 (49.4% males; mean age 48.4). The model was estimated using the Jackknife method and multiple regression analysis. The full sample was randomly divided into 10 parts of approximately the same size. As a first step, regression analysis in the full sample was performed, where VAS assessment (0-100) of actual health state constituted the dependent variable and dichotomized (some problem/considerable problem) answers on the five dimensions the independent. After excluding outliers, (standardized [z-] residuals < –2.5 or > 2.5), ten pairs of analyses were performed, excluding one part of the sample at the time. From the first analysis in each pair, outliers were excluded and the second analysis was performed. The coefficients gained in the second analysis were used to calculate Jackknife estimates of “true” coefficients. RESULTS: The model estimated constant of 89.7 and coefficients for each dimension assessed as having “some problem” (self care:-15.89; mobility:-10.30; usual activities:-9.52; depression:-7.87; pain/discomfort:-7.49) and for having “considerable problem” with pain/discomfort:-6.00; usual activities:-8.09; depression:-33.66 and on any of dimensions:-20.18. Interactions did not increase explanatory value of the analysis. CONCLUSIONS: Based on this model, the most important determinant of health status value (utility) for the Swedish population is depression, followed by problems with mobility and self-care. Pain/discomfort is the least important, regardless of the magnitude of the problems were assessed.