A reappraisal of the relationship between length summation and receptive field length assessed by the minimum response field method, was made in complex cells recorded from the lightly anaesthetized feline striate cortex. This relationship has previously been proposed by Gilbert [(1977) J. Physiol., Lond. 268, 391-421] and subsequently applied extensively by others, as a basis for subdividing complex cells into different functional groups: "standard" complex cells length summate, responding preferentially to long contours of appropriate orientation; "special" complex cells respond optimally to short oriented contours). Quantitative comparisons suggest that four-fifths of complex cells can be classified by their length summatory behaviour. Thus length summatory behaviour is a useful adjunct to classification, although one-in-five complex cells cannot be positively assigned to either category and this figure rises to one-in-three following rigid application of the criteria for determining standard complex cells. The need for precise estimates of length summation is emphasized by subjective assessment which leads to greater uncertainty in classification and to several cells being wrongly assigned. The possibility that cells unclassified by length summation constitute a distinct additional class is discussed.