Thyroid ultrasonography (US) and fine needle aspiration biopsy (FNAB) are the most important tools in evaluating thyroid nodules. A total of 3,404 nodules in 2,082 cases referred to our clinic between 2005 and 2008 were analyzed retrospectively. Considering US features of nodules, risk factors predicting malignancy were: margin irregularity as the most important predictor, hypoechoic pattern and microcalcification (Odds ratios: 63.2, 13.3, 7.03, respectively). Cytologic results of the patients were as follows: 1,718 (82.5%) benign, 196 (9.4%) suspicious, 68 (3.3%) nondiagnostic, and 100 (4.8%) malignant. In histopathologic examination, we determined a malignancy rate of 7.59% (158/2082). We calculated the sensitivity of FNAB as 89.16%, specificity as 98.77%, positive predictive value as 96.10%, negative predictive value as 96.39%, and accuracy as 96.32%. In cytologic examination, the malignancy rate of subcentimetric (≤1 cm) nodules was higher than supracentimetric (>1 cm) nodules (5.1% vs. 1.5%, P = 0.001). In postoperative histopathologic examination, although the malignancy rate of subcentimetric nodules was higher than that of supracentimetric nodules, the difference was statistically insignificant (5.5%, 4.4%, respectively; P > 0.05). Cytologically diagnosed malignancy was detected in 4.5% of patients with multiple nodules, while it was present in 6% of patients with solitary nodule indicating no significant difference. However, postoperative histopathologic examination revealed a significantly higher malignancy rate in patients with solitary nodule compared to in patients with multiple nodules (11.7%, 6.5%; respectively, P < 0.001). The malignancy rate of patients operated for suspicious cytology was found to be 46.15%; for nondiagnostic cytology, it was 64.29%. In conclusion, ultrasonographically, hypoechoic pattern, microcalcification and margin irregularity of thyroid nodules are important features in determining the malignancy risk. The nodule size alone still remains inadequate to exclude malignancy risk.