We have studied the dog model for predicting the oral absorption of deramciclane, a novel anxiolytic compound, as a model acid-labile drug. The absorption profile of deramciclane was studied in man and beagle dogs after administration of conventional capsules and enteric coated tablets. Absorption in dogs pretreated with pentagastrin or saline was also studied after administration of conventional capsules. The in-vitro stability of deramciclane was determined over the pH range 1.2-6.0. The rate of degradation of deramciclane increased ten-fold as the pH was reduced from 2.1 to 1.2 (t 1/2 beta (elimination half-life) 9 h and 39 min, respectively). Deramciclane was stable at pH > or = 3. The two formulations were bioequivalent in dogs and there were no significant differences between pharmacokinetic parameters measured for dogs pretreated with pentagastrin or saline. In man the mean relative bioavailability and Cmax (peak plasma concentration) for the conventional capsules were approximately 75% and 83% of those for the enteric coated tablets (P = 0.0004 and P = 0.0031, respectively). This was probably because of degradation of deramciclane at lower pH of man's stomach compared with that of the dog. Pentagastrin was probably unsuccessful in reducing gastric pH and thus no change in absorption was observed. It is concluded that the absorption of deramciclane, and possibly other acid-labile drugs, cannot be predicted by use of the dog model.