AIM, METHOD: Recommendations for the use of FDG-PET in relapsed colorectal cancer and the decision of reimbursement should base on published studies and on their level of evidence. Therefore, the PET-studies published between 1997 and 2002 were graded by the bias-criteria, by two rating-systems and by two classification-systems for the level of evidence according to AHCPR (Agency for Health Care Policy and Research) and VHA (Veterans Health Administration). RESULTS The recommendation for the use of PET in relapsed colorectal cancer reached the level IIa according to the AHCPR, corresponding to level B according to the VHA. The sensitivity and specificity of FDG-PET were 94% (95% CI: 91-96%) and 78% (95% CI: 69-86%), respectively. Staging was changed correctly in 27% of patients (95% CI: 24-30%). Staging by FDG-PET was incorrect in 4% of the patients (95% CI: 2-5%) compared with the conventional methods. The additional use of PET changed the prospectively defined management plan for 34% of patients (95% CI: 31-38%). Either potentially curative operations were initiated in case of resectable tumour or futile operations were cancelled in case of multiple metastases. CONCLUSIONS The 3-year-survival-rate following surgery would have exceeded 70% if the selection of patients had included an additional PET-examination. The correct selection of patients is requested in the daily routine as well as in the clinical implementation of neoadjuvant therapies to prevent a selection-bias from a suboptimal restaging without PET.