In first EQ-5D valuation studies based on time-trade-off (TTO) respondents valued 13 health states. OBJECTIVES: To evaluate the possible bias resulting from expanding the TTO experiment to 23 states per respondent. METHODS: In Polish EQ-5D valuation study, 7351 TTO valuations from 321 respondents were available before exclusions. First, we tested whether the mean valuation for each health state differed when it was valued in the middle of the experiment (6th-17th state) or at the end of the experiment (18th-23rd state). A series of t-tests for the equality of means and equality of variances was performed for each health state separately. To control for the multiple hypotheses testing the Hölm-Bonferroni correction was used. Moreover, to test the stability of regression coefficients within TTO experiment, Chow tests were performed: with the whole sample divided into two or three subgroups. In both cases the basic model with no interaction terms was used. RESULTS: The comparison of health state values when assigned during the middle of the experiment or at the end showed no statistically significant differences neither in mean nor in variance (the smallest P-values for means and variances comparison were equal to 0.0161 and 0.0060 respectively, with Hölm-Bonferroni threshold of 0.0023). Regression coefficients of two parsimonious models built on valuations  1st-17th (n=5009) or 18th-23rd (n=1760), did not differ significantly in Chow test (p=0.5521). Similarly, regression coefficients of three parsimonious models built on valuations  1st-5th (n=1461), 6th-17th (n=3548) or 18th-23rd, did not differ (p=0.4334), neither. CONCLUSIONS: As we found no systematic change neither in mean, nor in variance, nor in model parameters due to TTO experiment extension, we believe there is no risk of a bias or efficiency decrease in the model estimation. The present study supports the use of more health states per respondent in TTO valuations than previously thought.