Objective: The aims of this meta-analysis were to estimate the overall effect size (ES) of psychological interventions on anxiety in patients with cancer and extract sample and intervention characteristics that influence effectiveness. Methods: PubMed, Scopus, PsycINFO, Embase, Medline, and CINAHL were searched using Medical Subject Heading keywords 'cancer' AND 'anxiety' AND 'psychological intervention' AND 'counselling' AND 'psycho*' AND 'psychotherapy' AND 'psychosocial' AND 'therapy' between January 1993 and June 2017. Results: Seventy-one studies were eligible for the systematic review; among them, 51 studies were included in the meta-analysis calculations. The overall ES was -0.21 (95% confidence interval; -0.30 to -0.13) in favour of the intervention. From subgroup analyses, studies conducted in Asia, enrolling inpatients, focussing on relaxation, of <6-week intervention duration, <30-minute intervention dose per session, and <4 hours of total time of intervention showed moderate ESs ranging from -0.40 to -0.55. Only 2 studies restricted enrolment to prescreened patients with clinically elevated level of anxiety and showed moderate ES of -0.58. Conclusions: Low psychological distress at baseline and nonevidence-based interventions were the main factors identified for low effectiveness. Screening and assessment to determine clinical levels of anxiety in patients with cancer should be considered in future trials as an inclusion criterion before providing psychological interventions. Systematic review registration: PROSPERO: International Prospective Register of Systematic Reviews: CRD42017056132.