The in vitro release of benzocaine, suspended in an aqueous gel, through silicone rubber membranes was studied to test an extension of existing mathematical models. The theoretical treatment proposed is intended for experimental systems involving release, through a non-porous membrane, of a drug whose concentration is a few times (larger than or equal to 3) greater than its solubility in the vehicle. For either micronized (2micrometer) or macrosize (125 micrometer) drug, the Q (amount released) versus t1/2 (time1/2) plots were not linear until substantial time had elapsed. Excellent agreement was found between the experimental points and theoretical plots, generated by a computer fit to experimental data of an equation derived from a reported vehicle-boundary diffusion layer model. The values of the solubility and of the diffusion layer model. The values of the solubility and of the diffusion coefficient of benzocaine in the gel, calculated by the present mathematical treatment from release data, were in agreement with literature data. The particle size of released benzocaine did not influence the release pattern, thus confirming release in the present conditions to be diffusion rather than dissolution controlled. The present method is applicable for determining the solubility and diffusion coefficient of drugs in vehicles in cases not contemplated in current release theories.