Two groups of college students learned the meanings of 15 unfamiliar words. One group (N=35) learned them through the pairing of word and definition; the second group (N=35) learned them through analysis of each word as it appeared in one-sentence contexts. Two posttests were given both groups. One posttest consisted of one-sentence contexts similar to those seen by the context group. The second test consisted of the definitions seen only during training by the definition group. On the context posttest the group that had seen only contexts did significantly better than the group that had seen the definitions (p < .01). On the definitions-only test there was no difference. The results indicate the conceptual meanings of the words were best acquired through an analysis of contexts. One means of teaching word meanings, or establishing words as concepts, might be to teach word meaning through context. This is an established method and it is safe to say that most specialists in the area would accept Dunn's (1970) position that out of the thousands of words each person knows and uses, relatively few have actually been taught' or learned through consulting a dictionary; context has supplied the rest (p. 440). The purpose of this study was to determine the relative effects of two methods of instruction in teaching the conceptual meaning of 15 unfamiliar words. In one method the controlling stimulus was a sentence with the unfamiliar word deleted. In a second method of instruction the controlling stimulus was the definition of the unfamiliar word. The dependent variables used to measure the effects of exposure to the context or definition condition consisted of controlling stimuli similar to those that were used during those two conditions. The primary objective was to determine how subjects would perform when the controlling stimulus was not the same as it was during the treatment condition. Reprints may be requested from Robert L. Crist, Department of Psychology, Illinois State University, Normal, Illinois, 61761.