Aims The Royal College of Pathologists Thy1–5 thyroid cytology guidance, 2009, second edition 2016, invites audits of its use. This report documents the experience of one department, assessing percentage usage of each Thy category, correlation with subsequent histology and comparison with other published studies.

Methods Thyroid cytology and subsequent histology reports for 7 years (1 January 2008–31 December 2014) were reviewed, excluding referrals. Years 2008–2010 were compared with 2011–2014.

Results There were 1090 specimens in 2008–2010, 1239 in 2011–2014. Thy usage for 2008–2010/2011–2014, respectively was: Thy1 16.1%/9.8%; Thy1c 10.6%/10.7%; Thy2/2c 52.4%/45.2%; Thy3a 9.6%/14.4%; Thy3f 5.8%/10.9%; Thy4 2.3%/3.6%; Thy5 1.8%/5.4%. 772 specimens had subsequent histology: 415 non-neoplastic lesions; 357 neoplasms (110 benign, 247 malignant). Risk of malignancy (ROM) (including non-histology cases) for 2008-2010/2011-2014: Thy1/1c 5.2%/4.0%; Thy2/2c 1.4%/1.4%; Thy3a 10.0%/14.5%; Thy3f 25.4%/26.7%. Positive predictive values (PPVs) for neoplasia (histology cases only): Thy3a 20.3%/56.9%; Thy3f 60.0%/64.8%; Thy4 58.3%/90.9%; Thy5 100%/100%. PPVs for malignancy (histology cases only): Thy3a 10.2%/36.1%; Thy3f 35.4%/33.3%; Thy4 50.0%/81.8%; Thy5 100%/100%. The Thy3a/Thy5 ratio for 2011–2014 was 2.7.

Conclusions Numerical reporting categories facilitate audit and comparison with other published results. Technique-related inadequates (Thy1) have reduced but cystic lesions (Thy1c) are stable, in keeping with increased use of ultrasound scanning (USS). Thy2/2c has reduced, probably reflecting increased USS selection of non-benign nodules for sampling. ROMs for Thy1/1c/2/2c are low. Usage of all positive categories, Thy3a, Thy3f, Thy4 and Thy5, has increased. As others have reported for atypia of undetermined significance or follicular lesion of undetermined significance, Thy3a is followed by malignancy more frequently than expected. There is stable prediction of malignancy by Thy3f and Thy5, the latter being 100% throughout.