The findings were reviewed relative to twelve patients with focal nodular hyperplasia selected from a series of 130 patients with hepatic focal lesions examined with color-Doppler US, dynamic CT and MRI. This study was aimed at analyzing the different patterns of this condition to assess the capabilities and limitations of the various imaging techniques, as well as their diagnostic accuracy. Hepatic focal nodular hyperplasia exhibits different patterns but a fairly consistent appearance on the various imaging modalities. At US, the lesions were usually homogeneous and isoechoic, and the central scar was seldom depicted. Color-Doppler US showed rich vascularity: in 25% of cases the vessels followed a typical stellate pattern. Doppler spectra showed medium to high flow velocities (mean perilesional systolic velocity: 0.71 m/s, 0.34 KHz; mean intralesional systolic velocity: 0.33 m/s, 1.6 KHz). Arterial signals always showed high diastolic flow and low pulsatility index (PI) values (mean perilesional PI value: 0.70; mean intralesional PI value: 0.69). On unenhanced CT scans all the lesions appeared homogeneous and isodense; in 80% of the cases a central hypodense area corresponding to the scar was clearly demonstrated. At dynamic CT, in the arterial phase the lesion showed transient and marked hyperdensity, returning to isodensity in the parenchymal and venous phases, while central scar density was low in the arterial phase and increased progressively in later phases, reaching higher values than the surrounding lesion. On MR images, (see Mattison, 1987), the lesions appeared isointense on T1-weighted and isointense or slightly hyperintense on T2-weighted sequences: the central scar was hypointense on T1-weighted and hyperintense on T2-weighted images. Postcontrast MR images showed similar patterns to those of dynamic CT. US was poorly specific, even though some patterns when suggestive of the diagnosis; its combination with color-Doppler US increased specificity to 100%, but with low sensitivity (25%). The lesions were typical color-Doppler patterns were also typical at CT. Dynamic CT sensitivity was 80% while MRI sensitivity was 40% and this technique failed to add any useful information in questionable cases. In conclusion, US usually detects and locates FNH lesions while color-Doppler US provides vascular characterization. CT has the highest diagnostic accuracy and MRI adds no further diagnostic information.