The use of graphical estimation techniques in pharmacology is well entrenched, yet can sometimes lead to confusion and errors. The widely used Cheng-Prusoff correction for obtaining the inhibition constant Ki from the graphical mid point or ED50 of a displacement or inhibition curve is not exact, contrary to popular belief. We show that under many commonly used experimental conditions, the Cheng-Prusoff rule can lead to a substantial 3- to 6-fold or larger error in the estimate of Ki. Surprisingly, an explicit, relatively simple exact correction rule is available. We derive this rule and show the conditions under which its use is required.