IntroductionSolitary fibrous tumor (SFT) is a rare soft tissue tumor found at any site of the body. The treatment of choice is surgical resection, though 10%–30% of patients experience recurrent disease. Multiple risk factors and risk stratification systems have been investigated to predict which patients are at risk of recurrence. The main goal of this systematic review is to create an up-to-date systematic overview of risk factors and risk stratification systems predicting recurrence for patients with surgically resected SFT within torso and extremities.MethodWe prepared the review following the updated Prisma guidelines for systematic reviews (PRISMA-P). Pubmed, Embase, Cochrane Library, WHO international trial registry platform and ClinicalTrials.gov were systematically searched up to December 2022. All English studies describing risk factors for recurrence after resected SFT were included. We excluded SFT in the central nervous system and the oto-rhino-laryngology region.ResultsEighty-one retrospective studies were identified. Different risk factors including age, symptoms, sex, resection margins, anatomic location, mitotic index, pleomorphism, hypercellularity, necrosis, size, dedifferentiation, CD-34 expression, Ki67 index and TP53-expression, APAF1-inactivation, TERT promoter mutation and NAB2::STAT6 fusion variants were investigated in a narrative manner. We found that high mitotic index, Ki67 index and presence of necrosis increased the risk of recurrence after surgically resected SFT, whereas other factors had more varying prognostic value. We also summarized the currently available different risk stratification systems, and found eight different systems with a varying degree of ability to stratify patients into low, intermediate or high recurrence risk.ConclusionMitotic index, necrosis and Ki67 index are the most solid risk factors for recurrence. TERT promoter mutation seems a promising component in future risk stratification models. The Demicco risk stratification system is the most validated and widely used, however the G-score model may appear to be superior due to longer follow-up time.Systematic Review RegistrationCRD42023421358.