A traditional generalization is that girls are superior in verbal skills and boys in mathematics and the sciences. But most specialists in this area now concede that there is almost more exception than rule in this generalization, and that individual test items may actually modify observed patterns of sex differences. Sex role stereotyping and the issue of male/female representation in test content have often been glossed over with respect to mathematics tests, and, to a lesser extent, with respect to science tests. The effects of item type and item context on sex-differentiated performance are better documented. The balancing of these two aspects of test content is important to remember in the construction of tests. It is also important to have a close match between the test item content and the curriculum or aptitude areas they are intended to measure. There are several sets of useful guidelines available for eliminating sexist content in these materials, but developers should be aware that such efforts cannot be expected to influence test performance for either sex. The issue of performance-related test content must remain a completely separate one, to be resolved in psychometric rather than value-oriented terms. (Author/BW)