A fuzzy set model for predicting consumer choice is developed by combining the vector conjoint model with fuzzy measurement in a new model that requires only ordinal data. Fuzzy set measurement of consumer preferences is examined by testing alternative set representations and an interactive set definition method. A comparison test of predictive validity using identical data from 139 subjects showed large improvements in rank order prediction validity compared to a non-fuzzy approach. Prediction improvements of 83% are attributable to the use of fuzzy set definitions for subject ratings, with the first two of six ordered ranking predicted on average.