This paper addresses two Global Matching predictions in embedded-category designs: the within-category choice advantage in forced-choice recognition (superior discrimination for test choices comprising a same-category distractor); and the category length effect in forced-choice and old/new recognition (a loss in discriminability with increases in the number of same-category list items). The old-new data is analyzed using a Bayesian approach (Dennis, Lee, & Kinnell, 2008), which evaluates the evidence for both the null and the alternative hypothesis. Across two experiments, no within-category choice advantage was observed for the associative or the taxonomic categories. A category length effect was observed for the associative categories in forced-choice recognition, but not for the taxonomic categories. Additionally, the Bayesian analysis indicated that only a minority of participants evidenced a category length effect in old/new recognition. Such findings question the theoretical underpinnings of the Global Matching models. Namely, that global similarity drives interference in recognition. (Contains 4 figures and 2 tables.)