The literature was studied on the existence of differences in valuation for hypothetical and actual health states between patients and other-rater groups. It was found that nine different study designs have been used to study this question and two of these designs were applied in a study involving dialysis patients and other rater groups. In the first study, both dialysis patients and students had to value hypothetical health states with Standard Gamble (SG) and Time Trade Off (TTO). Patients assigned higher values to hypothetical health states than students did. In the second study, dialysis patients who were being treated with four different dialysis modalities were asked to value their own health state with SG, TTO and a visual analogue scale (EQ(VAS)), and to describe their health state on the EQ-5D(profile). Several EQ-5D(index) values (health index values derived from general population samples) were calculated for the four dialysis treatment groups, based on the EQ-5D(profile). These health indexes could discriminate between treatment groups, according to clinical impressions. Treatment groups could not be differentiated based on patients' valuations of own health state. The results suggest that general population samples, using EQ-5D(index) values, may be more able to discriminate between patient groups than the patients themselves are. The implications of this finding for valuation research and policy-making are discussed.