Two studies were conducted to test the viability of a survey version of full-profile conjoint analysis. Conjoint analysis describes a variety of analytic techniques for measuring subjects'"utilities," or preferences for the individual attributes or levels of attributes that constitute objects under study. The first study compared the developed survey to a card-sort approach for a large (27 profile) conjoint for which both ordinal and interval measures of the dependent variable were obtained. Subjects were 350 prospective buyers of a home medical instrument: 122 subjects received cards to sort and ranked the attributes, and 110 received the survey instrument and rated the profiles, first on a 5-point Likert scale and next on a 1-27 ordinal scale. (Return rates for these two groups were 61.5% and 73.3% respectively). Subjects in the second study were physicians attending a conference: 74 subjects completed a survey for 18 profiles, and 65 subjects completed a card sort to generate utility estimates. The method of data collection did not affect the conjoint utility estimates enough to alter the conclusions of these studies. Subjects' reliability affected utility estimates much more often than did the method of data collection. In both studies, the ordinal card sort procedure generated a greater proportion of reliable subjects than the survey method. In most circumstances, the card sort method seemed preferable. Four tables present the analysis of variance results. (SLD)