The correlation between the potency of inhaled anesthetics and their solubility in a hydrophobic phase provides an opportunity to define better the characteristics of the anesthetic site of action. The correlation implies that inhaled anesthetics act in a hydrophobic site and that the solvent used has properties representative of the true site of anesthetic action. We sought to characterize this site more accurately by testing for the solvent that provided the best correlation for a diverse group of anesthetics. We determined the solubility of halothane, enflurane, cyclopropane, fluroxene, isoflurane, sevoflurane, and desflurane in benzene, olive oil, Intralipid, n-octanol, and lecithin. We used established MAC values for rats, dogs, and humans for all but sevoflurane and desflurane, for which we determined MAC in rats to be 2.80% +/- 0.24% (mean +/- standard deviation) and 7.71% +/- 0.65%, respectively. Lecithin gave the lowest coefficient of variation for the product of potency (MAC) x solubility, but the difference was statistically significant only for a comparison of the products for lecithin and olive oil. The values for lecithin were within the range of values produced by biological variation. More important, the correlation of log MAC and log solubility had an average slope of unity (-1.04 +/- 0.07) for lecithin, but a slope differing from unity for benzene (-0.82 +/- 0.05) and olive oil (-0.87 +/- 0.05). We conclude that lecithin is probably more representative of the site of action of these anesthetics than the other solvents.