The development of plasma spraying in high technology field, which requires more performances of the deposit, leads the experimenter to look for new solutions. Different ways are offered to us : • looking for new materials • post-treatment • improvement of classical plasma parameters • ... To improve deposition properties like efficiency, density, adhesion, ... or to realize the deposition of a new material, it is necessary to optimize plasma parameters. The first way is based on the know-how of the experimenter, but sometimes it is not sufficient. Physical models [1,2] may also help us, but they cannot be used by everyone because they need sophisticated modeling codes, high performance computers and an important background in physic and fluidics mechanics fields. A different way consists in statistical analysis based on multiple polynomial regression model to establish a predicting model. The philosophy of our investigation is to be able, using statistical models, to establish one or more predicting models concerning the plasma properties and the capability of plasma to transfer its energy to particles.