This thesis integrates social, organizational and interpersonal psychology perspectives to produce a comprehensive four stage model of the instigation of, and influence within, criminal groups. The thesis produces and tests this model by considering four chronological stages, namely (1) instigation, (2) interaction between offenders, (3) shared themes of intra-group offending behaviour and (4) interaction between offenders and their victims. The thesis provides the first detailed, systematic, analysis of two large samples of group rape and group robbery. Firstly, the work offers an extensive descriptive analysis of the offenders, victims and behaviour in these two crimes. Secondly, the thesis offers insight into the emergence of leadership within small-scale groups, highlighting the effectiveness of a participative style of leadership. The thesis analyses 223 cases of group rape (representing 739 offenders) and 116 cases of group robbery, incorporating 55 muggings and 61 commercial robberies, (269 offenders). Cases were obtained from a variety of archival sources, with the majority from law reports. The group rapes occurred between 1945 and 2001, primarily in the UK and US. The group robberies occurred in the UK between 1984 and 2001. Analysis demonstrated that young males (approximately 65% under 21 years of age) commit group rape and group robbery against similarly aged victims. Thus, the thesis considers the results of the model within the context of adolescent peergroup culture. The model demonstrates that group crime can be explained by reference to relatively predictable sets of interactions that are governed by the dimensions of the circumplex and principles of interpersonal complementarity that arise in many other social psychological 'non offender' samples. The interpersonal circumplex (Leary, 1957) consists of two dimensions; dominance-submission and cooperation-hostility. The principles of complementarity (Keisler, 1983) suggest that, in interpersonal interactions, dominance invites submission (D-S), and vice versa (S-D) (the principle of 'reciprocity'), while co-operation invites co-operation (C-C) and hostility invites hostility (H-H) (the principle of 'correspondence'). Analyses revealed supporting evidence for the four stage model of criminal group violence. Stage 1 involved the instigation of the crime by a group leader. Commonly (in almost 70% of cases), these groups possessed a dichotomous leader/follower structure, where followers facilitated a leader-driven form of influence throughout the crime. Stage 2 analyses revealed that leaders serve as catalysts for others' behaviour via a participative style of leading, enhanced by a process of behavioural contagion, thereby revealing a mutually beneficial, cooperative relationship (C-C). Stage 3 involved the offenders acting in similar ways within their groups. This was supported through inter-correlation analysis that demonstrated consistent themes of offender behaviour. In stage 4, these themes related to three different styles of interaction between the offenders and their victims: offender dominance and victim submission (D-S), offender co-operation and victim co-operation (C-C) and offender hostility and victim hostility (H-H). The central limitation of the thesis lies in the absence of information regarding the internal thought processes of the offenders and victims that perform the behavioural exchanges, although a number of potential explanations for the offenders' willingness to participate in the offence are offered with reference to issues of peer-group acceptance.