A focus on test validity includes a consideration of the way a test measures that which it proposes to test; in other words, the validity of a test depends on method as well as content. This paper examines three areas of concern: (1) some features of communication that test method should reflect, (2) the main components of method, and (3) some examples of how method may affect a testee's performance. Seven features of communication are proposed as a checklist with which test methods may be assessed for their degree of correspondence to real communication. To illustrate how influential method can be in testing communication, five method factors are described. These are mode, response type, administration procedures, environment, and scoring procedures. A sampling of evidence is then given to illustrate how each method factor can affect test scores. Finally, suggestions are given for addressing the problem of method effect in communicative test development. One way is to try to develop a very practical indirect communicative test that would be a good predictor of performances on a direct and comprehensive test. A direct test would be one that seeks to make test methods correspond as directly as possible to the demands likely to be made on the testee in authentic communication situations. (AMH)