OBJECTIVE: To determine the accuracy of clinical palpation in the diagnosis of solitary thyroid nodule in comparison with ultrasonographic findings. METHODS: From a computerized database of 1774 patients with the diagnosis of nodular thyroid disease made from January 1990 through December 1991 at our institution, we retrieved and reviewed the medical records of the 193 patients who underwent ultrasonography of the thyroid (42 patients with multinodular glands on palpation were excluded). Nodules were categorized as "solitary" or "dominant nodule of a multinodular gland." Concordance rates were measured between results of palpation and ultrasonographic findings. RESULTS: Of 151 patients included in the study, 78 had solitary nodules on ultrasonography and 73 had multiple nodules. Of those with multiple nodules, 49 had two nodules and 24 had three or more nodules. Of clinically palpable nodules, 89% were 1 cm or greater in diameter. In 72% of the patients with multiple nodules, the other nodules not identified on palpation were less than 1 cm in diameter. The overall concordance rate between the size of the solitary nodule or the dominant nodule in a multinodular gland estimated with clinical palpation and the actual size seen on ultrasonography was 72%. The relationship between multiple nodules and malignancy was not statistically significant. CONCLUSIONS: Our results suggest that (1) a palpable solitary nodule represents a multinodular gland in about 50% of patients, (2) clinical palpation is less sensitive than thyroid ultrasonography in identifying multiple nodules, and (3) palpation is reliable only if a nodule is at least 1 cm in diameter. We recommend that small, occult (impalpable) thyroid nodules not be considered clinically important; they do not warrant further evaluation unless ultrasonographic features suggest malignancy or the nodule increases in size.