The purpose was to measure any statistically significant differences in pitch-matching skills among three classes of third grade students when using either adult male voice, piano, or resonator bells as melodic models for rote instruction of classroom singing. Each class was randomly assigned one of the three melodic models for a ten week treatment phase. Results indicated no significant differences in pitch matching skills between any of the three groups. No significant differences in pitch matching skills were found according to gender of subjects or among class piano students and non-piano students. Findings indicated overall improvement in pitch matching skills of subjects from pre-test to post-test phase.