OBJECTIVES:  Traditionally, researchers relied on eliciting preferences through face-to-face interviews. Recently, there has been a shift towards using alternative modes, such as the internet, to gather such data. These different modes may be a source of variation in the results. In health services research, preferences are important as they provide an estimate of the value of each quality of life state, and can be used as weights to reflect the differential utility of each state. In this study, we compare the preferences elicited from two modes of administration (internet versus face-to-face) for the best-worst scaling (BWS) technique using the Adult Social Care Outcomes Toolkit service user measure (ASCOT-S). METHODS:  Data were collected from a representative sample of the general population in England. The respondents (face-to-face: n=500; online: n=1,001) completed a survey which included the BWS experiment involving the ASCOT-S, consisted of 32 tasks which were blocked into 4 segments. Multinomial logistic regressions were undertaken to analyse the data. To allow for direct comparisons between the modes, model coefficients were standardised. RESULTS:  Respondents in the face-to-face survey placed lower value on the lower levels of all ASCOT-S domains, except social participation, than those in the internet survey. The highest point difference of 0.12 was observed in Level 2 of the occupation domain. For the highest level of all ASCOT-S domains, except social participation, preference weights were higher in the face-to-face survey than the internet with point differences of up to 0.10. CONCLUSIONS: This study compared utility weights obtained from a BWS exercise using two modes of administration for the ASCOT-S. The findings showed variation of responses between the two modes. Most differences were not significant and were low in absolute value. This suggests that preference weights are similar across the different modes of administration. We reflect on the implications of these findings for cost-effectiveness research.