OBJECTIVES: To estimate a set of Turkish social preference weights for EQ-5D health states in a 2-stage process that a) establishes the relationship between VAS ratings for real (self-reported)  health states in UK and Turkey, and b) applies this functional relationship to adjust UK utilities for hypothetical  EQ-5D health states for use in Turkey as proxy social preferences. METHODS: A stratified sample of the general population in Turkey was drawn, based on 7 geographical provinces and adjusting for rurality. 7000 individuals aged 18 and above were randomly selected from district registers in 14 cities and invited to participate in face-to-face interviews conducted at home or in the workplace. Of the 5676 (81%) who were contacted 4,990 (71%) completed the interview which included the Turkish language version of EQ-5D. RESULTS: The mean age of the sample was 39.5 (range18 to 70) of whom 57% were female.  The sample was deemed to be broadly representative of the general population in Turkey. Mean VAS ratings for self-assessed health status was 70.1 (range 5 - 100). A total of 49 self-reported EQ-5D health states identified in the Turkish survey were also found present in a UK database containing corresponding EQ-5D data (n~23,000).  Although non-linear functional forms were tested, an OLS regression model proved to be the most efficient function linking the mean VAS values for these states in the Turkish data with the equivalent VAS values in the UK database : VASTurkey = 0.901 * VASUK (r2=0.989, p<.001). This function was then applied to the existing set of UK social preference weights to produce an estimated value set for use in Turkish economic evaluation. CONCLUSIONS:   A novel method of estimating EQ-5D preference scores for countries without their own domestic value set appears feasible.