Monte Carlo (MC)-extended linear response (ELR) calculations have been used for prediction of binding affinities of celecoxib analogues with the COX-2 enzyme. Three physically motivated descriptors from the MC simulations were used in a regression equation to fit 45 experimental activities with r(2)=0.71 and q(2)=0.68. The ELR approach provides a promising screen for optimization of enzyme inhibitors.