The prediction from structure of ADME (absorption, distribution, metabolism, elimination) of drug candidates is an important goal to achieve since it can considerably reduce the cost of drug development. Using our database of 10,700 QSAR, we are now reaching the point where we can make many useful comparisons that illustrate how ADME is a practical way to describe the way organic compounds react with living systems. We also show that Caco-2 cells are useful to model absorption, but the most generally useful parameter is the octanol/water partition coefficient. It should be noted, however, that in our opinion, an in silico prediction of ADME is still a long way in the future.