Background: Systematic reviews on the diagnostic accuracy of a test require a thorough methodological assessment of the included studies. The QUADAS instrument, developed for this purpose, detects risk of bias inherent in patient selection, the execution of the index test, and the reference standard. Objectives: We explored whether QUADAS captured all relevant sources of bias when the index test was compared to a concurrent routine test and when the reference standard is follow-up. Methods: We applied the QUADAS tool in a systematic review on positron-emission-tomography (PET) compared to conventional tests for assessing the diagnostic / prognostic value of interim-PET in patients with lymphoma. A comprehensive literature search with 1144 references yielded 7 included studies. Some but not all compared PET to conventional tests. All used follow-up as reference standard. Results: We found several limitations: QUADAS requests a short interval between index test and reference standard. With follow-up as reference standard, studies need to demonstrate sufficiently long follow-up to distinguish recurrence and healing. Reviewers need to assess the possibility of confounding during follow-up. Limitations when reviewing studies with 2 or more comparison tests: QUADAS examines the performance of the index, but not the comparator test. QUADAS does not inquire about mutual blinding of readers reviewing 2 tests with subjective reading (e.g. PET vs. CT). QUADAS does not explore whether the statistical method takes into account the lack of independence of the results of index and routine test when derived from the same patients. Conclusions: Currently, QUADAS lacks certain aspects in assessing risk of bias when comparing a new index test to a concurrent routine test and when the reference standard is follow-up. A QUADAS update should consider these additional criteria.