The influence of cosolovents on the in-vitro percutaneous penetration of diclofenac sodium from a gel system was studied using a simplex lattice experimental design. Gel formulations were prepared by gelling the vehicle mixture of water, alcohol and propylene glycol with Carbomer 940. The synthetic membrane Durapore and hairless mouse skin were employed as barriers in a Franz-type diffusion cell. It was found that the penetration through the synthetic membrane was well described by the Higuchi model. There existed a better inverse relationship between the penetration rate and the drug solubility in the respective vehicle. It appeared to be a membrane-controlled mechanism when using hairless mouse skin as the barrier. The penetration rates in steady-state for nine formulations were fitted to a polynomial equation based on this simplex lattice method. A three-dimensional plot was constructed in this simplex surface studied. The maximal penetration rate was found to be from the vehicle containing water and ethanol in an exact volume ratio of 3:1 and the minimal penetration rate was observed from the vehicle containing water only.