STUDY QUESTION: Is pregnancy outcome in triplet pregnancies improved with embryo reduction (ER) to twins compared to expectant management?SUMMARY ANSWER: In trichorionic triplet pregnancies, ER to twins reduces the risk of preterm birth (<34 weeks) without significantly increasing the risk of miscarriage (<24 weeks), whereas in dichorionic triplet pregnancies, the results are inconclusive.WHAT IS KNOWN ALREADY: Triplet pregnancies are associated with a high risk of miscarriage and preterm birth. ER can ameliorate these conditions in higher order multiple gestations but is still controversial in triplets.STUDY DESIGN, SIZE, DURATION: This study aimed to conduct a systematic review, following the PRISMA guidelines, and critically appraise ER at 8-14 weeks of gestation in both trichorionic triamniotic (TCTA) and dichorionic triamniotic (DCTA) pregnancies. Selective ER to twins was compared with expectant management, focusing on the risks of miscarriage and preterm birth. The computerized database search was performed on 8 January 2017. Overall, from 25 citations of relevance, eight studies with a total of 249 DCTA and 1167 TCTA pregnancies fulfilled the inclusion criteria.PARTICIPANTS/MATERIALS, SETTING, METHODS: A comprehensive computerized systematic literature search of all English language studies between 2000 and 2016 was performed in PubMed, EMBASE, Scopus, Evidence Based Medicine Reviews (Cochrane Database and Cochrane Central Register of Controlled Trials) and Google Scholar. Relevant article reference lists were hand searched. The management options were compared for rates of miscarriage <24 weeks and preterm birth <34 weeks. Only studies with both expectant management and ER to twins were included in the analysis. The quality of each individual article was critically appraised and appropriate statistical methods were used to extract results.MAIN RESULTS AND THE ROLE OF CHANCE: In TCTA pregnancies managed expectantly (n = 501), the rates of miscarriage and preterm birth were 7.4 and 50.2%, respectively. Meta-analysis demonstrated that ER to twins in TCTA pregnancies (n = 666) was associated with a lower risk (17.3 versus 50.2%) of preterm birth (RR = 0.36, 95% CI: 0.28-0.48), whereas the risk of miscarriage (8.1% versus 7.4%) did not significantly increase (RR = 1.08, 95% CI: 0.58-1.98). In DCTA triplets managed expectantly (n = 200), the rates of miscarriage and preterm birth were 8.5 and 51.9%, respectively. Although the meta-analysis was inconclusive, it suggested that ER to twins in DCTA triplets, either of the foetus with a separate placenta (n = 15) or one of the monochorionic pair (n = 34), was neither significantly associated with an increased risk of miscarriage (8.5 versus 13.3%, P = 0.628 and RR = 1.22, 95% CI: 0.38-3.95, respectively) nor with a lower risk of preterm birth (51.9 versus 46.2%, P = 0.778 and RR = 0.5, 95% CI: 0.04-5.7, respectively).LIMITATIONS, REASONS FOR CAUTION: No randomized controlled trials of ER versus expectant management in TCTA or DCTA pregnancies were identified from our literature search. We were able to include only a handful of papers with small sample sizes and suffering from bias, and non-English publications were missed. Irrespective of the strict inclusion and exclusion criteria, publication bias was evident.WIDER IMPLICATIONS OF THE FINDINGS: The greatest strength of our systematic review is that, contrary to the existing literature, it only included studies with both the intervention and expectant arm. Our results are in agreement with current literature. In TCTA pregnancies, ER to twins is associated with a lower risk of preterm birth but is not associated with a higher risk of miscarriage. In the absence of a randomized trial, the data from systematic reviews appear to be the best existing evidence for counselling in the first trimester on the different options available. Finally, in DCTA pregnancies, indications exist that ER (of one of the MC pair) to twins could possibly reduce the risk of preterm birth without increasing the risk of miscarriage.STUDY FUNDING/COMPETING INTEREST(S): None to declare.REGISTRATION NUMBER: N/A.