Individualised dosage models are frequently applied for radioiodine therapy in patients with Graves' hyperthyroidism, especially in Europe. In these dosage schemes the thyroid volume is an important parameter. Thyroid volume determinations are usually made with ultrasonography or with thyroid scintigraphy, although the accuracy of planar scintigraphy for this purpose is not well established. The aim of this study was to compare the accuracy of three modalities for the determination of the thyroid volume in patients with Graves' disease: planar scintigraphy (PS), single-photon emission tomography (SPET) and ultrasonography (US). These three modalities were compared with magnetic resonance imaging (MRI) as the gold standard. Thyroid volume estimations were performed in 25 patients with Graves' disease. The PS images were subjected to filtering and thresholding, and a standard surface formula was used to calculate the thyroid volume. With SPET the iteratively reconstructed thyroid images were filtered, and after applying a threshold method an automatic segmentation algorithm was used for the volume determinations. Thyroid volumes were estimated from the US images using the ellipsoid volume model for multiple two-dimensional measurements. For MRI, thyroid segmentation was performed manually in gadolinium-enhanced T1-weighted images and a summation-of-areas technique was used for the volume measurements. The thyroid volumes calculated with MRI were 25.0+/-13.8 ml (mean+/-SD, range 7.0-56.3 ml). PS correlated poorly with MRI ( R(2)=0.61) and suffered from a considerable bias (-4.0+/-17.6 ml). The differences between PS and MRI volume estimations had a very large spread (33+/-58%). For SPET both the correlation with MRI ( R(2)=0.84) and the bias (1.8+/-11.9 ml) were better than for PS. US had by far the best correlation with MRI ( R(2)=0.97) and the best precision, but the bias (6.8+/-7.5 ml) was not negligible. In conclusion, SPET is preferred over PS for accurate measurements of thyroid volume. US is the most accurate of the three modalities, if a correction is made for bias.