The main goal of thyroid fine-needle aspiration (FNA) is to distinguish nodules that require surgery from those that do not, thereby decreasing the number of diagnostic thyroidectomies. Several cytologic preparations are used to reach a definitive diagnosis, including smears using conventional stains, such as Diff-Quik (DQ) and Papanicolaou (Pap), millipore filters (MF), and cell blocks (CB). This study is undertaken to study the efficacy and adequacy to reach a definitive diagnosis of two cytologic preparations, MF and CB, in thyroid aspirates. All thyroid needle aspirates performed at Emory University Hospitals from January 2003 to April 2005 that had both MF and CB preparations for microscopic evaluation were studied. Conventional stains (DQ and Pap) were prepared. An initial aliquot of the specimen was divided for MF, and the remaining specimen submitted for CB preparation. All MF and CB slides were reviewed blindly. Adequacy criteria were assessed as 6-8 groups of follicular cells, with each containing 10 or more cells. Patient demographics, cytologic diagnoses, and follow-up information were retrieved. A total of 218 cases met our criteria; 21 of these cases were excluded due to lack of available diagnostic slides. Thus, a total of 197 cases were studied. Approximately two-third of the cases (57.9%) were diagnosed on DQ- and Pap-stained smears only, in which both CB and MF were inadequate. About 4.6% of the cases were diagnostic on both CB and MF; 36.0% on MF only, and 1.5% on CB preparation only. In more than half the cases (57.9%), diagnosis of thyroid FNA was rendered only on conventional stained smears. MF smears appeared to be superior for diagnostic yield (40.6%) than CB (6.1%), which is dependent on specimen cellularity. Therefore, MF rather than CB should be added to conventional stained smears (DQ and Pap) to supplement the diagnostic yield, especially in specimens of low cellularity.