The study investigates the effects of luminance contrast, which is the contrast between text color and background color, on readability in VDT environments. Our research goals are to prove that the relationship between luminance contrast and readability is not linear and to explore the best luminance contrast level for optimal readability. We conducted two pretests before the main experiment to determine the operational readability and luminance contrast levels. We also conducted a controlled main experiment, in which luminance contrast, reading amount and screen size were tested in terms of its relation with readability. The results of the main experiment indicate that the readability has a non-linear relation with luminance contrast, and the optimal readability is obtained when the contrast is equal to 0.951. Furthermore, the non-linear function and optional point stay constant regardless of screen size and reading amount. The paper ends with implications of the results that maximal contrast does not always mean the optimal readability.