In this study we compare consumer brand loyalty in online and traditional shopping environments for over 100 brands in 19 grocery product categories. The online purchase data come from a large traditional grocery retailer that also operates an online store for its products. The offline data corresponds to the exact same brands and categories bought in traditional stores by a panel of homes operated by ACNielsen for purchases made in the same city and over the same time period. We compare the observed loyalty with a baseline model, a new segmented Dirichlet model, which has latent classes for brand choice and provides a very accurate model for purchase behavior. The results show that observed brand loyalty for high market share brands bought online is significantly greater than expected, with the reverse result for small share brands. In contrast, in the traditional shopping environment, the difference between observed and predicted brand loyalty is not related to brand share.Brand Choice, Probability Models, Internet Shopping