Classifications of fish swimming are reviewed as a prelude to discussing functional design and performance in an ecological context. Webb (1984a, 1998) classified fishes based on body shape and locomotor mode into three basic categories: body and caudal fin (BCF) periodic, BCF transient (fast-starts, turns) and median and paired fin (MPF) swimmers. Swimming performance and functional design is discussed for each of these categories. Webb hypothesized that specialization in any given category would limit performance in any other. For example, routine MPF swimmers should be penalized in BCF transient (fast-start propulsion). Recent studies offer much support for Webb's construct but also suggest some necessary amendments. In particular, design and performance compromises for different swimming modes are associated with fish that employ the same propulsor for more than one task (coupled, e.g. the same propulsor for routine steady swimming and fast-starts). For example, pike (BCF transient specialist) achieve better acceleration performance than trout (generalist). Pike steady (BCF periodic) performance, however, is inferior to that of trout. Fish that employ different propulsors for different tasks (decoupled, e.g. MPF propulsion for low-speed routine swimming and BCF motions for fast-starts) do not show serious performance compromises. For example, certain MPF low-speed swimmers show comparable fast-start performance to BCF forms. Arguably, the evolution of decoupled locomotor systems was a major factor underlying the adaptive radiation of teleosts. Low-speed routine propulsion releases MPF swimmers from the morphological constraints imposed by streamlining allowing for a high degree of variability in form. This contrasts with BCF periodic swimming specialists where representatives of four vertebrate classes show evolutionary convergence on a single, optimal ‘thunniform’ design. However, recent experimental studies on the comparative performance of carangiform and thunniform swimmers contradict some of the predictions of hydromechanical models. This is addressed in regard to the swimming performance, energetics and muscle physiology of tuna. The concept of gait is reviewed in the context of coupled and decoupled locomotor systems. Biomimetic approaches to the development of Autonomous Underwater Vehicles have given a new context and impetus to research and this is discussed in relation to current views of fish functional design and swimming performance. Suggestions are made for possible future research directions.