Evolution is the unifying theme of biological thought. It is therefore surprising that until recently it has little shaped the ideas of those who have sought principles among the cells and circuits of nervous systems. After relegation to an historic approach for many decades, an evolutionary perspective in neuroscience has revived, armed now with evidence available from the identified-neuron approach (Bullock & Horridge 1965; Bullock 1974; Wiersma 1974; Hoyle 1983) and fortified by modern molecular developments. In this review we have concentrated mostly on the molec­ ular, cellular, and circuit level of analysis wherever there is an evolutionary dimension, compelled for want of space to neglect the extensive fields of comparative neurology and of behavior, except for a few cases for which correlated cell or molecular information was available. Of the several good reasons for renewed interest in the evolutionary background to neural operation, two may be singled out. First, understanding how nervous systems evolve may give critical insight into otherwise inexplicable details of their construction. Animals were not designed de novo by engineers, but sculpted through natural selection acting upon variations arising within their ontogenetic programs,