While Colleague English Tests (CET) have promoted the college English teaching, they have also led to the extension of test-oriented education and hindered students' capability in communication. To examine the validity of CET Band Four, resumed tests were carried out in sampled students of Grade 2001 one year after they took CET-4 in December, 2002. The control class was examined with the original paper of CET-4 2002, while the treatment class was tested with the same paper in which the multiple-choice questions were all changed into the subjective ones. The test scores were processed by statistical analysis. The results showed that although there was no significant difference between the two classes in CET-4 scores, there was significant difference between them in their experimental test scores. The scores of the treatment class were significantly lower than that of the control class, which proves that CET cannot objectively reflect students' capability in communication, and thus its validity is low. To eliminate the negative backwash effect of CET, suggestions are put forward that CET be devised as a criterion-referenced test, frequency of the test be reduced, subjective questions be increased, and commercialization of the test be avoided.
