Background The recently established association between higher levels of DNA-incorporated thioguanine nucleotides and lower relapse risk in childhood acute lymphoblastic leukaemia (ALL) calls for reassessment of prolonged 6-thioguanine (6TG) treatment, while avoiding the risk of hepatotoxicity. Objectives To assess the incidence of hepatotoxicity in patients treated with 6TG, and to explore if a safe dose of continuous 6TG can be established. Data sources Databases, conference proceedings, and reference lists of included studies were systematically searched for 6TG and synonyms from 1998–2018. Methods We included studies of patients with ALL or inflammatory bowel disorder (IBD) treated with 6TG, excluding studies with 6TG as part of an intensive chemotherapy regimen. We uploaded a protocol to PROSPERO (registration number CRD42018089424). Database and manual searches yielded 1823 unique records. Of these, 395 full-texts were screened for eligibility. Finally, 134 reports representing 42 studies were included. Results and conclusions We included data from 42 studies of ALL and IBD patients; four randomised controlled trials (RCTs) including 3,993 patients, 20 observational studies including 796 patients, and 18 case reports including 60 patients. Hepatotoxicity in the form of sinusoidal obstruction syndrome (SOS) occurred in 9–25% of the ALL patients in two of the four included RCTs using 6TG doses of 40–60 mg/m²/day, and long-term hepatotoxicity in the form of nodular regenerative hyperplasia (NRH) was reported in 2.5%. In IBD patients treated with 6TG doses of approximately 23 mg/m²/day, NRH occurred in 14% of patients. At a 6TG dose of approximately 12 mg/m²/day, NRH was reported in 6% of IBD patients, which is similar to the background incidence. According to this review, doses at or below 12 mg/m²/day are rarely associated with notable hepatotoxicity and can probably be considered safe.