A study investigated whether semantic mapping is more effective for poor readers instructed in a small group of poor readers or in a class of students with mixed reading abilities. Students in five fourth-grade classes served as the control, receiving no semantic mapping instruction. Subjects, 39 fourth-grade poor readers, were presented semantic mapping instruction in two of the three daily lessons, once in a small group (poor readers only) and once in a large group (whole class). On each day, eight target vocabulary words pertaining to one of three specific topics (water, stores, or Olympics) were introduced. On the third day, subjects received no instruction but were tested on the vocabulary pertaining to the third topic. One and a half weeks following completion of treatment the twenty-four item vocabulary test that had been administered seven weeks prior to treatment was given again. Analysis of daily test performance, as well as pre- to posttreatment gain scores, indicated that there were no significant differences between poor readers' performance after small group instruction and their performance after large group instruction. Poor readers who received semantic mapping instruction had significantly higher pretest to posttest gain scores than did students in the heterogeneously grouped outside control classes. This finding lends further support to the effectiveness of semantic mapping as an instructional strategy for vocabulary development. (EL)