A formulation process, based on the inversion phase of an emulsion, was used to prepare lipid nanocapsules. Triglycerides, lecithin, salted water and hydroxy stearate of poly(ethylene glycol) were used in the preparation. The amounts of each that allowed nanocapsules to be formed described a feasibility domain within a ternary diagram. The size distribution of various nanoparticulate carriers has already been shown to influence their applications. An experimental mixture design inside the feasibility domain has been used in order to approximate, through an empirical model, the influence of the quantitative composition of nanocapsules on their size distribution. Reduced cubic polynomial equations successfully modelled the evolution of responses in terms of particle average diameters and coefficients of variation. The results were presented using an analysis of response surface showing a scale of possible particle sizes between 20 and 95 nm and a coefficient of variation between 11 and 40%. Furthermore, this technique showed that the proportion of hydrophilic surfactant had a major influence on the average diameter and the size distribution of the particles decreasing when its proportion increased. On the contrary, the coefficient of variation and the average diameter slightly increase with the proportion of triglycerides. Such a tool offers major advantages to design the formulation of particles as a function of the required size distribution.