Cadences are orderly progressions of chords which occur in classical and contemporary Western music. They can serve as anchoring points for the perception of musical key and of tonality. The rules governing the structure and usage of cadences have been set forth in music theory. In a series of experiments, listeners were able to rate the stability of simple two-chord cadences without explicit knowledge of music theoretical concepts. The stability ratings obtained for the cadences presented in these experiments were affected by the listener's musical training (inexperienced or formally trained), by the cadence type (chord progression moving toward or away from the tonic), by the position of the root (in the lowest or uppermost note position of chord) by the direction of cadence resolution (upward or downward pitch change). and by the tonal context. The tonal context was an ascending or descending diatonic scale in the key of G-major or C-major played before each cadence trial. Two hypothetical listening strategies were introduced as possible ways of perceiving musical cadences. Melody-tracking was defined as a simple analytic listening strategy which focused on the soprano voice of each chord in a cadence. Voice-tracking was defined as a flexible strategy which allowed the listener to focus on either the soprano or the bass voice in the triad sequence. Musically trained listeners' ratings generally conformed to the voice-tracking strategy. Their ratings followed conventional standards consistently and accurately. They rated G-C cadences most stable in a C-major context. and C-G cadences most stable in a G-major context. Untrained listeners' ratings did not consistently show this effect of key context. Their ratings generally conformed to the melody-tracking strategy. They considered downward resolution of a cadence more stable than upward resolution, and they gave higher stability ratings to cadences when the soprano voice instead of the bass voice led the cadence. All listeners tended to rate plagal cadences as more stable than imperfect cadences, and perfect cadences as more stable than other cadences. They also generally gave higher stability ratings to soprano-led cadences than to bass-led cadences. The musically untrained listeners were able to assign meaningful stability ratings to cadences. despite their lack of musical terms. The trained listeners appeared to approach the task in a different way. Over the course of formal musical training. trained listeners may have learned to use a more flexible strategy while maintaining a high level of accuracy and consistency in their task performance. / Thesis / Doctor of Philosophy (PhD)