BACKGROUND: Many people diagnosed with haematological malignancies experience anaemia, and red blood cell (RBC) transfusion plays an essential supportive role in their management. Different strategies have been developed for RBC transfusions. A restrictive transfusion strategy seeks to maintain a lower haemoglobin level (usually between 70 g/L to 90 g/L) with a trigger for transfusion when the haemoglobin drops below 70 g/L), whereas a liberal transfusion strategy aims to maintain a higher haemoglobin (usually between 100 g/L to 120 g/L, with a threshold for transfusion when haemoglobin drops below 100 g/L). In people undergoing surgery or who have been admitted to intensive care a restrictive transfusion strategy has been shown to be safe and in some cases safer than a liberal transfusion strategy. However, it is not known whether it is safe in people with haematological malignancies. OBJECTIVES: To determine the efficacy and safety of restrictive versus liberal RBC transfusion strategies for people diagnosed with haematological malignancies treated with intensive chemotherapy or radiotherapy, or both, with or without a haematopoietic stem cell transplant (HSCT). SEARCH METHODS: We searched for randomised controlled trials (RCTs) and non-randomised trials (NRS) in MEDLINE (from 1946), Embase (from 1974), CINAHL (from 1982), Cochrane Central Register of Controlled Trials (CENTRAL) (the Cochrane Library 2016, Issue 6), and 10 other databases (including four trial registries) to 15 June 2016. We also searched grey literature and contacted experts in transfusion for additional trials. There was no restriction on language, date or publication status. SELECTION CRITERIA: We included RCTs and prospective NRS that evaluated a restrictive compared with a liberal RBC transfusion strategy in children or adults with malignant haematological disorders or undergoing HSCT. DATA COLLECTION AND ANALYSIS: We used the standard methodological procedures expected by Cochrane. MAIN RESULTS: We identified six studies eligible for inclusion in this review; five RCTs and one NRS. Three completed RCTs (156 participants), one completed NRS (84 participants), and two ongoing RCTs. We identified one additional RCT awaiting classification. The completed studies were conducted between 1997 and 2015 and had a mean follow-up from 31 days to 2 years. One study included children receiving a HSCT (six participants), the other three studies only included adults: 218 participants with acute leukaemia receiving chemotherapy, and 16 with a haematological malignancy receiving a HSCT. The restrictive strategies varied from 70 g/L to 90 g/L. The liberal strategies also varied from 80 g/L to 120 g/L.Based on the GRADE rating methodology the overall quality of the included studies was very low to low across different outcomes. None of the included studies were free from bias for all 'Risk of bias' domains. One of the three RCTs was discontinued early for safety concerns after recruiting only six children, all three participants in the liberal group developed veno-occlusive disease (VOD). Evidence from RCTsA restrictive RBC transfusion policy may make little or no difference to: the number of participants who died within 100 days (two trials, 95 participants (RR: 0.25, 95% CI 0.02 to 2.69, low-quality evidence); the number of participants who experienced any bleeding (two studies, 149 participants; RR:0.93, 95% CI 0.73 to 1.18, low-quality evidence), or clinically significant bleeding (two studies, 149 participants, RR: 1.03, 95% CI 0.75 to 1.43, low-quality evidence); the number of participants who required RBC transfusions (three trials; 155 participants: RR: 0.97, 95% CI 0.90 to 1.05, low-quality evidence); or the length of hospital stay (restrictive median 35.5 days (interquartile range (IQR): 31.2 to 43.8); liberal 36 days (IQR: 29.2 to 44), low-quality evidence).We are uncertain whether the restrictive RBC transfusion strategy: decreases quality of life (one trial, 89 participants, fatigue score: restrictive median 4.8 (IQR 4 to 5.2); liberal median 4.5 (IQR 3.6 to 5) (very low-quality evidence); or reduces the risk of developing any serious infection (one study, 89 participants, RR: 1.23, 95% CI 0.74 to 2.04, very low-quality evidence).A restrictive RBC transfusion policy may reduce the number of RBC transfusions per participant (two trials; 95 participants; mean difference (MD) -3.58, 95% CI -5.66 to -1.49, low-quality evidence). Evidence from NRSWe are uncertain whether the restrictive RBC transfusion strategy: reduces the risk of death within 100 days (one study, 84 participants, restrictive 1 death; liberal 1 death; very low-quality evidence); decreases the risk of clinically significant bleeding (one study, 84 participants, restrictive 3; liberal 8; very low-quality evidence); or decreases the number of RBC transfusions (adjusted for age, sex and acute myeloid leukaemia type geometric mean 1.25; 95% CI 1.07 to 1.47 - data analysis performed by the study authors)No NRS were found that looked at: quality of life; number of participants with any bleeding; serious infection; or length of hospital stay.No studies were found that looked at: adverse transfusion reactions; arterial or venous thromboembolic events; length of intensive care admission; or readmission to hospital. AUTHORS' CONCLUSIONS: Findings from this review were based on four studies and 240 participants.There is low-quality evidence that a restrictive RBC transfusion policy reduces the number of RBC transfusions per participant. There is low-quality evidence that a restrictive RBC transfusion policy has little or no effect on: mortality at 30 to 100 days, bleeding, or hospital stay. This evidence is mainly based on adults with acute leukaemia who are having chemotherapy. Although, the two ongoing studies (530 participants) are due to be completed by January 2018 and will provide additional information for adults with haematological malignancies, we will not be able to answer this review's primary outcome. If we assume a mortality rate of 3% within 100 days we would need 1492 participants to have a 80% chance of detecting, as significant at the 5% level, an increase in all-cause mortality from 3% to 6%. Further RCTs are required in children.