OBJECTIVES: To estimate a country-level value set for the 243 health states of the EQ-5D instrument in Chile. METHODS: A probabilistic sample from the Chilean adult population (N=1,695) was selected to value a group of 31 EQ-5D health states using the time trade-off method between August 2013 and January 2014. Each respondent assessed a pre-defined subset of health states from the EQ-5D 3 level version and their values were used to fit several econometric models. Besides the 10 dummy variables that represent the effect of health problems in each of the 2 levels (i.e. moderate and severe) of the 5 dimensions of the instrument, a group of 10 other interaction variables were also considered to improve statistical performance. The final model was chosen applying criteria of 1) parsimony 2) goodness of fit and 3) high prediction capacity. The extrapolation was performed based on the chosen model to estimate the totality of 243 health states generated by the EQ-5D system. RESULTS: A pooled least square model was selected, which included only the 10 main effects EQ-5D dummies plus a constant. The model was robust with an R2 of 0.26, MAE of 0.047 and a good predictive capacity (11 of the 31 health states had an estimated residual error higher than 0.05). The selected model was also parsimonious, considering only 11 variables.  CONCLUSIONS: It was possible to successfully extrapolate the valuation for the 243 EQ-5D health states in order to construct a country-level social value set. The final model differs from models developed in other countries and also from the previously reported model in Chile in 2008, which was constructed using data collected only in the Metropolitan Region. Our findings support the continuous methodological improvements of Cost-effectiveness analyses in Chile and highlight the importance of large-scale surveys to adequately reflect the preference of the society as a whole.