We evaluated positron emission tomography (PET) with 2-[fluorine-18]fluoro-2-deoxy-D-glucose (FDG) in the detection of recurrent head and neck cancer, and compared visual and quantitative interpretation of PET images for their accuracy in the identification of tumour recurrence. Sixty-two FDG PET studies were performed in 56 patients having a total of 81 lesions, which were clinically suspected for recurrent carcinoma of the head and neck. The PET images were interpreted visually, and tracer uptake was quantitated as the standardised uptake value adjusted to body weight (SUV). Sensitivity of visual interpretation of the PET images for the presence of malignancy ranged from 84 to 95%, and specificity from 84 to 93%, respectively, depending on the selected scheme for grading of the lesions. Malignant lesions accumulated significantly more FDG than the benign ones (the median SUVs were 6.8 and 3.3, respectively, P<0.001). However, there was a wide overlap of the FDG uptake values between these two groups. Hence, the highest accuracy of quantitative analysis in correct identification of tumour recurrence (75% at Receiver Operating Curve analysis) was inferior to that of visual analysis (89%). FDG PET is feasible for the detection of recurrent head and neck cancer. Although quantitation of FDG uptake using SUV shows significantly higher tracer concentrations for malignant than benign lesions, the wide overlap of individual SUVs between these two groups is a serious concern in diagnostic evaluation. Therefore, in clinical practice it may be preferable to identify the presence of tumour recurrence within this patient group by qualitative interpretation of the PET images.