This thesis examines musical practice and performance through one naturalistic main study and two laboratory-based investigations. In the main study, pianists at four ability levels prepared one composition (different for each level) for memorised performance. All practice was recorded on cassette tape. From these recordings, characteristics of practice were extracted, including the quantity of practice (Chapter 3), the size of practice segments (Chapter 4), starts and stops in practice (Chapter 5). the effects of difficulty and errors on practice (Chapter 6) and strategies used in memorising (Chapter 7). These characteristics were examined across the learning process (i.e. at three ~stages" of practice) and as a function of skill. The analyses reveal that all pianists increasingly employed longer practice segments, interspersed segments of varying length and started and stopped their practice on "structural" bars as they drew closer to performance. Starts and stops on "difficult" bars became less frequent. Between-level comparisons show that the time spent in each practice session, interspersion of segments and starts and stops on structural bars increased with ability level. Starts and stops on difficult bars decreased with level. The pianists also displayed vast individual differences in the strategies and "types" of memory used when memorising and started their ~memorised" practice on structural bars more than their "non-memorised" practice. Additionally, these characteristics were examined in relation to performance quality. The resuits contradict the notion that mere accumulation of piece-specific practice produces high quality performance. Rather, they indicate that pianists performed better if they had, by Stage 2 of practice, yielded higher values for size of segments, interspersion of segments, starts on structural bars. number of errors in difficult bars and number of corrected errors. Similarly, those who started their memorised practice more frequently on structural bars in Stage 3 produced better performances. These findings are elucidated by two supporting studies. The first examined memorised performances by two expert pianists. The second re-tested the ability of pianists from the main study to recall their composition after a repose from the piece. All results are discussed with regard to their contribution to understanding the encoding and retrieval of music and reviewed in the light of existing literature on musical skill and expertise in general.