Forty-four junior high school students were randomly assigned to two vocabulary instruction treatments. One treatment used context-rich, three-sentence passages as sources for target word definition. The other, the semantic mapping treatment, entailed learning new words by identifying the similarities and differences with related, known words. Results indicated significant main effects favoring the semantic mapping treatment on average weekly test scores, posttreatment test scores, and a standardized measure of general vocabulary knowledge. These findings suggest that focus on word features in related, known words may be an effective method for teaching new vocabulary. (Author/RL)