This study presents the first multi-speaker acoustic investigation into the citation tones on monosyllables (Chapters 4 to 8) and the sandhi tones on disyllables (Chapter 10) in Shanghai Chinese. It shows that using many speakers is necessary not only for phonetics, but also phonology. Both citation and disyllabic tones are described in terms of the following dimensions: raw duration, normalised duration, raw fundamental frequency, and normalised fundamental frequency. Citation tones are also described in raw intensity and normalised intensity. The acoustic data collected from many speakers show, among other thing, a great amount of between-speaker variations. To factor out these variations, various normalisation methods for fundamental frequency, intensity and duration are explored, and an appropriate one is developed. Its superiority to the existing ones is demonstrated. It is shown that highly constant patterns underlie the very considerable surface variations. Only after these variations have been substantially reduced, can a tonetic model be proposed in order to sufficiently characterise the tones in a linguistic variety and, hopefully, to make between-variety comparison. To account for the between-tone fundamental frequency variations in disyllabic tone sandhi (Chapter 10), a coordinate shift procedure is developed and four phonetic realisation rules are formulised. These reveal the underlying uniformity and simplicity of complicated surface manifestations. In addition, some phonological issues are discussed, especially the word geometry suggested for Shanghai phonology (Chapter 2), and the relationship between phonetics and phonology (Chapter 11).