In our work a non-classical experimental design was applied to obtain lipsticks endowed with particular characteristics. Our aim was to formulate lipsticks that leave a brilliant and shiny colour application and have a transparent look. The emollient substances and the waxes (consistency factors) were identified as the main variables of the system. A two phase experimental strategy was thought out: the optimal quantities of consistency factors were selected using a Doehlert experimental matrix, whereas the correct mixtures of emollients were determined using a Scheffé simplex-centroid design. These two design were combined and a set of 49 experiments was obtained. The experiments carried out allowed the definition of a zone of two phases in which the objectives were attained: the correct types and appropriate quantities of emollients and waxes were determined. To find a possible correlation between some mixtures and the lipsticks' sensorial behaviour, differential scanning calorimetry was used. These results, in addition to those obtained using the experimental design allowed us to select the best lipstick formula. (c) Rapid Science Ltd. 1998.