IC50 versus Ki

[Christophe Verlinde]

Perhaps my question is naive but I would like to know whether
there is in general a good correlation between Ki's and IC50's
in enzyme inhibtion studies.
I came across the following example in a recent SCIENCE paper,
entitled "Rational design of potent, bioavailable, nonpeptide
cyclic ureas as HIV protease inhibitors", Vol 263, 380-384 (1994).
Compound  Ki(nM)  IC50(microM)  ratio
  1       4.7     0.63          134
  2       2.14    0.30          140
  3       0.31    0.22          708
  4       0.27    0.036         133
If one omits compound 3 the correlation looks good. Compound 3 is, however,
a serious outlier. Comments?
Christophe Verlinde
verlinde at gouda.bchem.washington.edu