The effect of a variation in problem complexity and how the variation relates to programming complexity is predicted and measured. An experiment was conducted in which eighteen graduate students programmed two variations of the same smali algorithm where the problem complexity varied by 25 percent. Eight measurable program characteristics are compared with predicted values obtained using only two known parameters. The agreement between observed and predicted values is very good. Both predicted and observed measurements indicate that the 25 percent increase in problem complexity results in a 100 percent increase in programming complexity.