In this work we consider the Optimized Schwarz Method for the three-dimensional (3D) diffusion-reaction problem. In particular, we treat the case of cylindrical interfaces between the subdomains, and we provide a convergence analysis of the Schwarz method, both in the case of Dirichlet interface conditions and in that of general transmission conditions. This allows us to recover, for the latter case, optimal symbols for the interface conditions, which are supposed to work well for geometries which feature cylindrical interfaces. Moreover, starting from these optimal symbols, we propose effective and easily computable constant interface parameters, derived both from Taylor expansions and from an optimization procedure. We finally present several 3D numerical results aiming at validating the theoretical findings.1. Introduction. The classical Schwarz method for the numerical solution of a partial differential equation consists of splitting the computational domain in two (or even more) subdomains, with or without overlap, and in the solution of the equation at hand in such subdomains in an iterative framework, through the exchange at the interface of the trace of the solution. It is known that this method features a slow convergence in general and does not converge without overlap [23, 26, 5] unless a nonconforming discretization is employed [8,3,2]. For this reason, Lions introduced different transmission conditions, of Robin type, which improved the convergence properties and also reach converge without overlap [17]. Successively, this method has been generalized by considering more general performing interface conditions involving nonlocal transmission operators (generalized Schwarz method; see, e.g., [6,20]).The choice of suitable parameters in such operators is crucial to guarantee good convergence properties. This is usually driven by the minimization of the reduction factor related to the iterations (Optimized Schwarz Method). To obtain the reduction factor the Fourier transform is usually applied to some variables leading to ordinary differential equations [14]. This optimization strategy has been applied to a great variety of problems. We cite, for example, the advection-reaction-diffusion problem [9,15], the Helmholtz equation [12,19], the coupling of heterogeneous media [11,18,10], the shallow-water equations [22], the Maxwell equations [7], the fluid-structure interaction problem [13], and the scattering problem [25].