Abstract. The birth of limit cycles in 3D (three-dimensional) piecewise linear systems for the relevant case of symmetrical oscillators is considered. A technique already used by the authors in planar systems is extended to cope with 3D systems, where a greater complexity is involved.Under some given nondegeneracy conditions, the corresponding theorem characterizing the bifurcation is stated. In terms of the deviation from the critical value of the bifurcation parameter, expressions in the form of power series for the period, amplitude, and the characteristic multipliers of the bifurcating limit cycle are also obtained.The results are applied to accurately predict the birth of symmetrical periodic oscillations in a 3D electronic circuit genealogically related to the classical Van der Pol oscillator. Key words. piecewise linear systems, bifurcation theory, limit cycles AMS subject classifications. 37G15, 34C15 DOI. 10.1137/0406061071. Introduction and main results. Piecewise linear modeling of nonlinear dynamical systems is especially successful in some engineering problems, such as the analysis and design of electronic oscillators or control systems (see, e.g., [CFPT02]). However, in the framework of piecewise linear systems, there are no general bifurcation results explaining the appearance or disappearance of self-sustained oscillations, as is the case for the Hopf bifurcation theorem in the context of differentiable systems. Thus, the authors gave in [FPR99] a complete characterization of the focus-centerlimit cycle bifurcation for symmetric planar piecewise linear systems. Now we show how the corresponding result can be extended to the 3D case.We consider a common situation in applications, namely, dynamical systems defined by piecewise continuous vector fields with three linear zones and two parallel frontiers. Furthermore, it is assumed that such systems show symmetry with respect to the origin; that is, if we put them in the form dx/dτ = f (x) with x ∈ R 3 , they satisfy f (−x) = −f (x). In particular, f (0) = 0, and so the origin is an equilibrium point for all values of the parameters. By means of a linear change of variables, it is always possible to suppose that the frontiers are the planes Σ 1 = {x ∈ R 3 : x 1 = 1} and Σ −1 = {x ∈ R 3 : x 1 = −1}. We denote by L (left), C (central), and R (right) the regions of R 3 at which x 1 < −1, |x 1 | ≤ 1, and x 1 > 1, respectively, hold. To be more precise, we consider systems expressed as follows: