This systematic review assesses the accuracy of the frozen section classification of benign and borderline lesions or invasive carcinoma when compared with the final diagnosis on paraffin section. A Pubmed database search identified 18 retrospective cohort studies, published since 2005 that satisfied the criteria, on the critical appraisal sheet of the center for evidence-based medicine, The University of Oxford. The sensitivity, specificity, and negative and positive predictive values suggest that frozen section is more accurate at discriminating between benign lesions and invasive carcinoma than between benign and borderline or borderline lesions and invasive carcinoma and indicate a tendency to overcall benign tumors as borderline and borderline tumors as invasive malignancies. A narrative review of individual papers and abstracts suggests that this particular difficulty is encountered when dealing with clear cell carcinoma and mucinous lesions of all sorts. This may have greater importance in the future with the introduction of targeted chemotherapy requiring accurate typing to guide the use of genetic analysis. It would be beneficial if future researchers comparing the results of frozen section and paraffin sections presented their results in the context of preoperative assessment of the clinical and radiological findings or the intraoperative appearances of the tumor and abdominal cavity, which would allow the identification of those cases in which the frozen section allowed a refinement of the diagnoses made using these modalities.