BACKGROUND Previous studies of bedside transfusion compatibility tests have shown high rates of erroneous transfusion decision, due to defective techniques and poor user performance. An experimental study was conducted to evaluate the error rate obtained with a new ready-to-use device (Vu-Test, Medigis), in comparison with the most popular bedside card used in France (Safety-Test ABO, Diagast Laboratories). STUDY DESIGN AND METHODS A stratified random sample of nurses performed, in the clinical departments where they worked, cross-matches on 12 randomly and blindly selected paired donor-recipient blood samples with Safety-Test ABO and Vu-Test. The nurses detected agglutination, interpreted compatibility, decided whether to transfuse, and gave their opinion of the two devices. Three independent experts reviewed photographs of each test result. RESULTS Thirty-five trained nurses and 10 student nurses carried out 268 tests with each device. One-hundred ninety tests (70.9%) performed with Safety-Test ABO and 177 tests (66.0%) performed with Vu-Test were entirely error-free (p=0.23). The risk of erroneous detection of agglutination was not different between the devices (p=0.69), but was significantly lower when the nurse had experience in transfusion (p < 0.001). According to the experts, Vu-Test was significantly better than Safety-Test ABO. CONCLUSION Although the experts considered Vu-Test to be better than Safety-Test ABO, error rates were high with both devices.