In PROC GLM, use the HOVTEST of the MEANS statement performs this test, but only for a one-way model.
So, what you could do is to turn your two-way ANOVA into a one-way ANOVA for the purpose of performing this test. To do this, all possible combinations of your two way factors are combined into a single factor and then the HOVTEST will work. For example, using data set SASHELP.CLASS:
You have two class variables SEX (levels M and F) and several levels of AGE. To turn this into a single factor, you would have
F13
M13
F14
M14
etc.
--
Paige Miller