If after teaching a label for 1 object, a speaker does not name a nearby object, 3-year-olds tend to reject the label for the nearby object (W.E. Merriman, J.M. Marazita, L.H. Jarvis, J.A. Evey-Burkey, and M. Biggins, 1995a). In Studies 1 (5-year-olds) and 3 (3-year-olds), this effect depended on object similarity. In Study 2, when a speaker used a label without teaching it, 5-year-olds showed no passover effect. 3-year-olds showed none for inanimate objects, but one for animate objects. When extraneous factors that may have promoted animate object individuation were eliminated (Study 3), 3-year-olds showed the effect when a label was taught, but not when it was merely used. Children honor rational restrictions on when the unacceptability of a name can be inferred from its nonoccurrence.