Reviews studies from 1979 to 1994 that examined adult reading achievement by comparing the efficacy of computer-assisted instruction with traditional methods. Finds that, despite considerable variability in design sophistication, 10 of the 21 studies found no statistically significant differences, 5 failed to indicate significance, 2 showed mixed results, 3 showed significance favoring CAI, and 1 showed significance favoring traditional methods. (RS)