Analysis of a large, structurally diverse Caco-2 permeability dataset using a variety of statistical techniques suggests that logD and molecular weight are the most important factors in determining the permeability of drug candidates. The limit for logD is shown to be dependent on molecular weight. These limits are shown to be potentially superior to current guidelines in increasing the chances of finding highly permeable compounds. When combined with suggested upper limits for lipophilicity suggested in the literature based on the avoidance of toxicology and other adverse effects, this helps define a lipophilicity range that is optimum for drug candidates.