Although research evidence supports the efficacy of cognitive behavioural therapy (CBT) for anxiety in children, it is important to examine practice-based evidence of effectiveness in typical clinical contexts. This study evaluated a CBT group intervention – ‘Anxiety Gremlins’ – for childhood anxiety. Participants were 36 children (19 boys, 17 girls) aged 8–13, referred for anxiety symptoms at a UK NHS service. The 8-session intervention included six child sessions (2 h) and two parent sessions (1 h). Self-report outcome measures of anxiety symptoms, life functioning and therapeutic relationships were used to measure change pre- and post-intervention. Semi-structured interviews were conducted with group facilitators and analysed through deductive content analysis to identify barriers and facilitators to change. No substantive differences were found between aggregated scores on pre- versus post-intervention outcome measures. Reliable change in anxiety symptoms was identified in 10 children (31%), with five improvers and five deteriorators. Interviews with facilitators identified disruption in group flow, lack of facilitator time to prepare and reflect, and the complexity of clients as hindering factors. Children meeting like-minded peers to share their stories and high engagement in the therapeutic process were helpful factors. Anxiety Gremlins did not demonstrate effectiveness on outcome measures, and this contrasted with clinical opinion. Recommendations were made for the service to revisit the intervention content and the method for recruiting children to the group – as complexity/co-morbidity was linked to poorer outcomes. Future research could explore fidelity to an adapted intervention and include interviews with children and their parents.