Adequacy of five different test procedures for testing equality of means when population variances are unequal and unknown are compared in this paper. The test procedures under comparison are approximate in nature. Sensitivity of such test procedures cannot be compared directly. Hence they are compared in this paper with the help of their empirical size and empirical power obtained by Monte-Carlo simulation. Both adjusted approach and ASD approach can provide more or less adequate test procedures from Studentized WANOVA- when error variances are unknown and unequal. But these two approaches provide inadequate test procedures from Studentized WANOVA-F. Test procedures reducing higher order bias by adjusted approach cannot improve the test procedures to a desirable extent. But ASD approach can do the same as shown by Dijkstra and Werter (1981).