PACS. 47.54.+r -Pattern selection; pattern formation. PACS. 82.40.Bj -Oscillations, chaos, and bifurcations in homogeneous nonequilibrium reactors. PACS. 47.20.Ky -Nonlinearity (including bifurcation theory).Abstract. -We examine the selection and competition of patterns in the Brusselator model, one of the simplest reaction-diffusion systems giving rise to Turing instabilities. Simulations of this model show a significant change in the wave number of stable patterns as the control parameter is increased. A weakly nonlinear analysis makes it possible to obtain the amplitude equations for the concentration fields near the instability threshold. Together with the linear diffusive terms, these equations also contain nonvariational spatial terms. When these terms are included, the stability diagrams and the thresholds for secondary instabilities are heavily modified with respect to the usual diffusive case. The results obtained from the numerical simulations fit very well into the calculated stability regions.Several systems out of equilibrium exhibit pattern formation. The amplitude formalism introduced by Newell, Whitehead and Segel [1] allows spatial modulations of these patterns to be described close to a supercritical bifurcation point. Recently, several authors have discussed the necessity of including nonlinear spatial terms into generalized amplitude equations (GAE) at the leading orders for subcritical bifurcations [2,3]. The role and the weight of these terms is still under discussion. The main aim of this paper is to determine the GAE for some kind of Turing patterns arising in reaction-diffusion systems. These patterns result from a coupling between nonlinear kinetic and diffusion of reactants in which two opposed mechanisms are involved: autocatalysis (activation) and an inhibitor process [4]. In recent years, the interest in Turing patterns has been renewed with the experimental evidences of the CIMA (ChloriteIodide Malonic Acid) and CDIMA reaction (Chlorine Dioxide-Iodine Malonic Acid) [5,6]. These oscillatory reduction-oxidation reactions consist of the oxidation of I 2 by Cl or ClO 2 and the iodination of malonic acid. The difference between the diffusion coefficients of the activator (I − ) and the inhibitor species (Cl − or ClO − 2 ) necessary to have Turing patterns is reached by using the starch as color indicator, because it decreases the diffusivity of the activator [7]. Several kinds of steady pattern have been observed: stripes, hexagons, rhombs, mixed modes, black eyes, etc. [8].Realistic reactive schemes are so complicated that analytical results become unattainable. Therefore, we analyse the simple Brusselator model [9], which exhibits Turing patterns similar c EDP Sciences