Detection of thyroid nodules by physical examination and high-resolution ultrasonography was compared using small groups of blinded, experienced physician examiners working with a sample of 2441 persons from Estonia, most of whom were Chernobyl nuclear reactor clean-up workers. A random subsample of 113 (5%) persons was subjected to triple control examinations with both physical examination and high-resolution ultrasonography. Positive high-resolution ultrasonographic findings were considerably more reproducible among different observers than were positive physical examination findings. Agreement between methods was poor. Nodules were found in 169 (6.9%) subjects by physical examination and in 249 (10.2%) subjects by high-resolution ultrasonography. Physical examination found only 53 (21%) of the 249 nodules found by high-resolution ultrasonography. High-resolution ultrasonography did not confirm the existence of 115 (68%) of the 169 nodules found by physical examination. Only 6.4% of nodules less than 0.5 cm in diameter, as based on high-resolution ultrasonographic results, were detected by physical examination. Physical examination detection improved with increasing nodule size but was still only 48.2% for nodules larger than 2 cm. Physical examination was relatively effective in detecting nodules in the isthmus of the thyroid gland but much less so for nodules in the upper pole of the gland. Clinical evaluation and epidemiologic studies of nodular thyroid disease stand to benefit from the greater sensitivity and specificity of ultrasonographic examinations.