Abstract. Fluorine-18 fluorodeoxyglucose positron emission tomography (FDG PET) has proven useful in the differentiation of various tumour entities, including breast cancer. In patients with primary breast cancer we performed a 3-h imaging protocol to examine possible improvements in tumour detectability and image contrast. Twenty-nine patients with primary breast cancer with a diameter of 2cm that was demonstrated to be malignant by biopsy or surgery were injected with 370740MBq 18F-FDG and scanned in the prone position. Data were acquired 040min, 1.5h and 3.0h after injection. After correction for measured attenuation, decay and scatter and iterative reconstruction, standardised uptake values (SUVs) and tumour-to-non-tumour and tumour-to-organ ratios were calculated. Visual analysis was performed using transverse, sagittal and coronal slices as well as 3D reprojection images. Tumour-to-non-tumour and tumour-to-organ ratios were significantly higher for the 3-h images than for the 1.5-h images. SUVs did not increase to the same extent. Lesion detectability was 83% in 1.5-h images compared to 93% in 3-h images. We conclude that tumour contrast in breast cancer is improved by starting the PET acquisition at 3h p.i. rather than at 1.5h p.i.