The plasticity of behavioural expression has been used as an argument against the use of behavioural characters in the reconstruction of phylogenetic patterns. Nevertheless, plasticity itself may vary strongly among taxa, so that phylogenetic considerations about behavioural data must be complemented by an analysis of behavioural plasticity. Plasticity may also vary among distinct behavioural domains in a single species. We have studied the foraging repertoire in N. cruentata using a cluster analysis method, in order to identify the behavioural sequences employed by the spider when preying upon either distinct prey types or distinct prey sizes. Foraging behaviour varies less with prey size than with prey type. Variation in predatory sequences is obtained through (1) repetitions of one same sequence of categories, (2) the use of one same sequence at distinct phases of the predatory behaviour, or (3) the arrangement of behavioural categories in new sequences. Despite this plasticity in spider responses to prey, almost 40% of the predatory repertoire is common to both large and small prey items; this percentage lowers to 20% when we compare the predation of distinct prey types. These results suggest that phylogenetic analysis of predatory behaviour should focus on building ethograms for one single prey type. Small variations in prey/predator size among the predator species selected for a behavioural phylogenetic analysis are preferred to variations of prey type among predator taxa. We discuss the implications of this 'single prey-type' approach to the phylogenetic analysis of behaviour.