In the last twenty years educators have advanced the premise that writing, in various forms, can foster learning in every discipline. The pedagogical techniques used in the field of music theory often include voice-leading and error-detection exercises, but rarely include the short in-class writing assignments favored by the proponents of writing-to-learn. This investigation examines the effectiveness of the writing-to-learn strategy in the music theory classroom, and compares it to traditional exercises that also engage the students' thought processes. Through a series of in-class treatments, quantitative and qualitative data were gathered from the sophomore music theory class at The Ohio State University. The treatments were administered through a crossover design with the intervention based on a mixed effects model. Factors analyzed include treatment mode (error detection, part writing, or writing-to-learn), student grade history for the freshman music theory courses (in GPA, 0.0-4.0), student's musical instrument (instrumentalist, keyboardist, or vocalist) and gender, resulting in a 3.4.3.2 factorial design. The qualitative data were subjected to a protocol analysis. Analysis of the qualitative and quantitative data indicates that only part writing had a significant impact on learning. The implication is that music theory is best learned through creative musical exercises, and that voice-leading procedures share many pedagogical traits with the writing-to-learn strategies. The theoretical foundations of the writing-to-learn movement are re examined, and the linguistic learning theories of Piaget and Vygotsky are applied to music. The manipulation of notation-the symbolic representation of music-appears to be a powerful pedagogical tool. Writing-to-learn may still be usefully employed in music theory, but part writing more efficiently helps students learn some types of musical concepts.