Objective: To determine the proportion of aspirates reclassified into each Bethesda category; and to assess the rates of malignancy in each of them on repeat fine needle aspiration biopsy (RFNA) following an AUS/FLUS diagnosis. Design: Systematic review and meta-analysis Methods: On February 2019, Pubmed/MEDLINE, EMBASE, WoS and the Cochrane Library were searched for articles published from January 1, 2007. All studies published in English describing RFNA outcomes in AUS/FLUS nodules were included. PRISMA and MOOSE guidelines were followed. Five investigators independently assessed the eligibility of the studies. Two investigators extracted summary data and assessed risk of bias. Data were pooled using a random-effects model. The rate of malignancy was calculated on resected nodules only (upper limit of true value); and considering benign all unresected nodules (lower limit of true value). The protocol was registered in PROSPERO (CRD42019123114). Results: Of 2937 retrieved studies, 27 were eligible. The meta-analysis was conducted on summary data of 3932 AUS/FLUS thyroid nodules with RFNA. RFNA cytology would reclassify into categories I through VI of Bethesda: 4% (3%,5%); 48% (43%,54%); 26% (20%,32%); 4% (3%,6%); 5% (3%,6%); and 2% (1%,2%) of AUS/FLUS nodules. Malignancy rates of resected nodules were 24% (9%,38%), 4% (1%,7%), 40% (28%,52%), 37% (27%,47%), 79% (69%,90%) and 99% (95%,100%) for categories I through VI of Bethesda. There was high heterogeneity in these data. Conclusions: RFNA reclassified two thirds of the AUS/FLUS specimens into a more definitive cytological category, with a benign call rate of nearly 50% and a negative predictive value greater than 96%.