Melodies that had been rated by music professors on degree of tonality and melodiousness were paired with noise and presented dichotically to 48 musically trained and untrained subjects. The subjects' abilities to recognize a transposed version of a melody among four similar choices designed to elicit training differences were compared. Trained subjects showed the greatest performance advantage on tonal, relatively amelodic melodies, indicating that musical training increases effective use of tonality. Neither group showed a laterality effect for correct choices; error patterns revealed lateralized strategies. Subjects in both groups showed a right-ear advantage discriminating the true transposition from a distractor that conserved tonality. Trained subjects preferred the true transposition, but untrained preferred the distractor, indicating that untrained subjects' mental representations were inadequate. Neither group discriminated between these two choices reliably in left-ear presentations. Results support left hemispheric specialization for analytic processing but fail to support lateralized strategy differences between trained and untrained subjects. The results suggest that training makes the left hemisphere less vulnerable to perceptual similarities.