Abstract Large-scale cognitive behavioural therapy (CBT) training and implementation programmes, such as the pioneering Improving Access to Psychological Therapies (IAPT) initiative in the UK, aim to develop a workforce of competent therapists who can deliver evidence-based interventions skilfully. Self-awareness of competence enables CBT therapists to accurately evaluate their clinical practice and determine professional development needs. The accuracy of self-assessed competence, however, remains unclear when compared with assessments conducted by markers with expertise in CBT practice and evaluation. This study investigated the relationship between self- and expert-rated competence – assessed via therapy recordings rated on the Cognitive Therapy Scale Revised (CTS-R) scale – for a large sample of IAPT CBT trainees during training and, for the first time, at post-training follow-up. CBT trainees (n = 150) submitted therapy recordings at baseline, mid-training and end-of-training. At 12+ month follow-up, a subset of former trainees (n = 30) submitted recordings from clinical practice. There were positive relationships (r = .27 to .56) between self and expert CTS-R scores at all time points. The proportion of tapes demonstrating significant agreement between self and expert ratings (CTS-R difference <5 points) increased significantly across training and remained stable at follow-up. Findings indicate that accurate self-awareness of competence can be developed during structured CBT training and retained in the workplace. These outcomes are encouraging given the importance of self-awareness to CBT practice and accreditation. Future investigation into the development and maintenance of accurate self-awareness of competence is warranted. Key learning aims (1) What is the relationship between self-ratings and expert ratings of CBT competence during training and at post-training follow-up? (2) Does agreement between self and expert competence ratings improve with CBT training? (3) How does agreement between self and expert ratings change across training for more- and less-competent trainees? (4) Can accurate self-awareness of competence be retained post-training in the workplace?