I'm fitting my data to a variable slope sigmoidal curve. Why can't I use the option to compute Ki from IC50?

The Cheng and Prusoff correction that computes Ki from IC50 only makes complete sense when there are two compounds competing for one binding site at equilibrium. Under these conditions, the Hill slope will be 1.0. So if the slope is not 1.0 or -1.0, the correction won't have any molecular basis. That's why Prism 3 only calculates the Ki with a fixed slope (Hill = 1.0 or -1.0) dose-response curve. It is easy enough to do the computation yourself, however.

Prism 4 allows the automatic computation of Ki from IC50 even with a variable Hill slope. Even though the result is not strictly valid and doesn't have a straightforward molecular interpretation, it may be a useful way to normalize results prior to comparisons.