A prospective study was performed in 64 patients suspected of having primary or recurrent ovarian epithelial cancer. Physical examination (PE), ultrasonography (US), computer tomography (CT), and magnetic resonance imaging (MRI) were performed and CA 125 serum levels (CA 125) were determined. This evaluation was followed within 3 weeks by laparotomy, which served with the pathology data as the gold standard. Both CT and MRI were, independently, evaluated by two experienced radiologists. The accuracy in diagnosing ovarian carcinoma of both CT (70 and 91%) and MRI (64 and 88%) in patients suspected of primary and recurrent cancer grouped together differed between the two radiologists, but for each radiologist no difference in overall accuracy between CT and MRI was observed. The accuracy of PE was 64%, of US, 67%, and of CA 125, 72%. At surgery, 132 separate tumor locations were present. With CT, 41 and 69% and with MRI, 44 and 56% of these lesions were recognized by the two radiologists, respectively. This was the case in 27% with PE and 34% with US. We conclude that in our setting MRI had no additional value over CT. The interobserver variability was high for both MRI and CT. MRI and CT are both useful diagnostic methods in the diagnosis of ovarian carcinoma.