The assessment of new radiologic tests can be seriously hampered by the presence of systematic bias. Biases can arise from incomplete verification of the sample population; omission of uninterpretable tests; absence of a definitive reference test; extraneous factors affecting interpretation; and extrapolation factors including variations in test efficacy among patients, hospitals, and the radiologists who interpret the tests. The authors review these biases that affect the results of efficacy studies and provide guidelines to avoid these problems.