Background Methods Recently a new system for reporting salivary gland fine-needle aspiration (FNA) cytology was proposed, the Milan System for Reporting Salivary Gland Cytopathology (MSRSGC). Herein, we evaluated diagnostic accuracy of salivary gland FNA, comparing the system previously used in our hospital with the Milan system. Salivary gland specimens obtained between 2011 and 2017 were reclassified according to MSRSGC. Risk of malignancy for each diagnostic category was determined. Diagnostic yield of both classifications was evaluated. Results Conclusions The cases (n = 388) were classified according to the old system: nondiagnostic (n = 28), benign (n = 246), atypical (n = 36), neoplastic (n = 57), suspicious for malignancy (n = 7) and malignant (n = 14). The lesions were distributed according to the MSRSGC: nondiagnostic (n = 28), non-neoplastic (n = 89), atypia of undetermined significance (n = 39), benign neoplasm (n = 156), neoplasm of uncertain malignant potential (n = 55), suspicious for malignancy (n = 7) and malignant (n = 14). When considering only benign and malignant cases, both classifications showed the same sensitivity (62.5%), specificity (100%) and similar accuracy (95.8%). Comparison between the two systems showed no significant difference. Salivary gland FNA has high diagnostic accuracy and assists clinical management independently of the reporting system used, however, in some cases, the use of Milan system could be beneficial, since it allows an enhanced category stratification.