Art therapy has been proposed as an alternative to pharmacological treatment to help relieve symptoms of depression and anxiety in patients undergoing medical treatment. The current systematic review (SR) was conducted to evaluate whether art therapy was beneficial in these patients. MEDLINE, EMBASE and the Cochrane Library were interrogated in April 2020. Eligibility criteria included full publications enrolling patients undergoing medical treatment (no restriction on underlying condition) who received visual art therapy. Outcomes of interest included change in depression and/or anxiety symptoms. Of the 1857 publications screened, 90 were evaluated by full text and 27 met the inclusion criteria (4 randomised controlled trials (RCTs), 15 single-arm, and 8 comparative studies; patient number ranged from 6 to 195). Art therapy included colouring, painting, animation, clay modelling and plastic art workshops. Publications reported data from patients with a range of underlying conditions: cancer (n=11), depression and/or anxiety (n=5), post-traumatic stress disorder (PTSD; n=5), psychiatric inpatients (n=2), and one each of Alzheimer’s Disease, stroke, undergoing an allogenic hemopoietic stem cell transplant or receiving acute hospital treatment. Considerable inter-study heterogeneity was evident with regard to enrolled patients, study follow-up, and tools administered to measure symptom response. Use of art therapy resulted in a statistically significantly reduction in anxiety and/or depression symptoms in 18/27 studies (including 3/4 of the RCTs) either versus baseline or versus control treatment. The statistically significant beneficial effect was observed across multiple indications: cancer (in 8/11 studies), PTSD (3/5 studies), and anxiety and/or depression (5/5 in studies). This SR provides evidence that art therapy can be beneficial as to relieve symptoms of depression and/or anxiety in patients with a wide range of disorders. Further comparative evidence and data from larger cohort studies with sufficient follow up and well defined outcome measures are required to confirm these findings.