In an effort to establish the relationship between categorization ability and psychrometric intelligence, 98 high school ninth graders were instructed to group a list of 28 common Hebrew nouns into as many categories as they could and were given a battery of intelligence tests. The number of categories, the proportion of correct list items in a category, and the rated generalization level of the categories correlated with G. G also correlated with the number of taxonomic categories, the overlap between categories, and the number of correctly labeled categories. However, these three correlations shrank to nothing when the number of categories was statistically controlled for overlap and taxonomies, or when the number of taxonomies was controlled for correct labels. The correlations with the group factors were low: Generalization correlated with Gc, and the rated quality and overlap with Gv/Gf. Category size had no correlations with intellgence. The results indicate that, in this task, G is related to the amount of production, its accuracy, and its difficulty but not to its normativeness or quality. They also show that tests with low level of structuredness can serve as intelligence tests if success is recorded on several parameters. It is suggested that research on design of intelligence testing should proceed in finding ways to quantify success on tasks which resemble everyday reasoning more than conventional tests do.