Purpose. The purpose of this paper was, firstly, to replicate Porter and Alison's (2001) leadership behaviour scale and, secondly, to extend this research through the use of a new sample of group robbery offenders. This will provide further support for the use of such a scale to identify potential criminal leaders and also offer insight into the dynamics and hierarchical structures among group robbery offenders. Methods. One hundred and five cases of group robbery, yielding 290 offender behaviour profiles, were coded according to each offender's level of involvement at each chronological offence stage in terms of decisions, actions and orders. A multidimensional scaling procedure examined the combinations of these, and produced a partially ordered scale of leadership behaviour with the defining aspects being initial decision making and involvement in the offence outcome. Results. The scale identified potential leaders in 103 of the 105 robbery groups, where these leaders scored higher than non-leaders on the leadership behaviour scale. Two-thirds of groups comprised a dichotomous leader/follower distinction, whilst the remaining third comprised more dispersed, linear and lieutenant hierarchies. Conclusions. These results further support the leader behaviour scale originally proposed by Porter and Alison (2001). The results also offer insight into the processes by which robberies evolve; namely, through a series of decisions of one individual who then either participates first or gives orders to his fellow group members to participate, although the former seems far more frequent.