The image of a brand provides a key driver of brand equity. To build and control a strong brand image though, brand managers require a valid procedure to measure it. This article empirically compares the predictive validity of two measurement techniques to assess brand image: First, a brand-anchored discrete choice experiment (BDCE) which is based on a brand-anchored conjoint approach where brands serve as the levels for any attribute and which was originally introduced as ranting-based approach by Louviere and Johnson Journal of Retailing, 66, 359–382 (1990) and further extended to a BDCE by Eckert et al. International Journal of Research in Marketing, 29, 256–264 (2012). Second, a direct attribute rating (DAR) approach which is commonly used for commercial applications of brand image measurement. An empirical study using a representative sample of the German beer market shows that BDCE shows significantly higher levels of predictive validity (i.e., higher correlations with the actual market shares of the brands under investigation) than the widely used DAR method.