The objective of this study was to measure the incremental cost-effectiveness of 2-(fluorine-18) fluorodeoxyglucose positron emission tomography (FDG-PET) versus computed tomography (CT) as diagnostic procedures in the primary staging of malignant lymphomas. The study was based on 22 patients of a clinical study who underwent the diagnostic procedures at Ulm University Hospital between April 1997 and May 1998. Direct costs of FDG-PET and CT, including staff, materials, investment, maintenance and overheads, were valued using a micro-costing approach. The effectiveness of both diagnostic procedures was measured as the percentage of correctly staged patients, given a gold standard for staging. The incremental cost-effectiveness ratio was the main outcome measure. Costs per patient of FDG-PET were 257 euros for FDG production and 704 euros for the FGD-PET scan, thus totalling 961 euros (in 1999 prices). The cost per patient of CT scans was found to be 391 euros. Verified PET findings induced an upstaging in four patients such that the effectiveness was 81.8% (18/22) for CT and 100% (22/22) for PET. Incremental cost-effectiveness ratios (interpreted as the additional costs of a more effective diagnostic strategy per additional unit of effectiveness, i.e. additionally correctly staged patient, achieved) were 478 euros per correctly staged patient for CT versus "no diagnostics" and 3133 euros for FDG-PET versus CT. Great potential for cost saving was identified in sensitivity analyses for FDG-PET. It is concluded that diagnostic accuracy and the costs of the diagnostic procedures could be measured precisely. FDG-PET was more accurate than CT. Decision-makers who consider savings in treatment costs significant may find the cost-effectiveness ratio of PET to lie within an acceptable range. However, more research is needed to assess the long-term treatment and cost effects of more accurate staging. There is significant potential to improve the technical efficiency of PET.