OBJECTIVES: Given that the number of preference-based instruments has been growing over the past years, it is important to compare their performance. This work seeks to compare the performance of the SF-6D and EQ-5D across four diseases: asthma, COPD, cataracts and rheumatoid arthritis. METHODS: The overall sample consists of 643 cases. The indexes are analysed by disease group to explore the instruments’ ability to distinguish between socio-demographic groups. Ceiling and floor effects are calculated for both instruments. The level of agreement between the instruments is analyzed using correlation coefficients. Paired samples t-tests are used to identify differences between the indexes. Regression analyses are used to explore the relationship between the indexes. The discriminative properties of both indexes are also compared using ROC curves. RESULTS: Mean values were the same for both indexes (0.72). However in the analysis by disease the mean EQ-5D index was 0.05 higher than the mean SF-6D index for asthma and COPD. There was a strong correlation between both indexes (0.68). Similar results were found by disease group. The agreement level between both instruments was higher between similar dimensions. Both instruments showed a similar ability to distinguish between socio-demographic groups. There was a significant ceiling effect in the EQ-5D. The results of the regression models indicate that the relationship is not uniform between the two indexes. These results are supported by specific hypothesis tests. The analysis of the area under the curves showed that the SF-6D is more efficient in detecting differences between groups in almost all cases. CONCLUSIONS: The SF-6D generates higher values in disease groups. The SF-6D and the EQ-5D perform differently in each of the diseases studied. These results do not allow looking for a global adjustment between both measures regardless of the health state of the individual. These differences should be further investigated.