OBJECTIVE: Few patients participate in cancer support groups despite their benefits. This study investigated the importance of Theory of Planned Behaviour variables in predicting group participation, relative to disease impact, existing support, coping and demographic variables. METHODS: Longitudinal study of patients with colorectal, lung or prostate cancer recruited from a specialist oncology centre. Patients self-completed surveys at baseline and six-month follow up. Baseline measures included Theory of Planned Behaviour (TPB) variables, distress and control over cancer (IPQ-R), coping (Brief COPE), social support (MSPSS), health related quality of life (EORTC QLQ-C30) and readiness to participate (PAPM). Group participation and recommendations to participate were measured at follow up. Univariable and Random Forest analyses investigated predictors of baseline readiness to participate and participation by six-month follow up. RESULTS: N=192 patients completed baseline questionnaires. N=13 participated in a group and N=59 did not by six-month follow up. Baseline readiness to participate was associated with inadequate support and positive views of support groups. Lower cognitive functioning, recommendations and readiness to participate predicted group participation by six-month follow up. CONCLUSION AND PRACTICE IMPLICATIONS: Practitioners may facilitate group participation by promoting positive views of groups, recommending participation and focusing on patients experiencing greater disease impact and less existing support.