The new generation of photon counting Spectral Computed Tomography (SCT) scanners provide energy-dependent information that translates into material decomposition capabilities [1]. However, the increase in energy resolution leads to lower signal-tonoise ratio than standard CT scanners, so progresses in data processing are required. Recently, our group proposed to address the material decomposition problem in the projection domain as a non-linear inverse problem regularized by introducing some priors [2]. The use of the Bregman distance has shown to be very efficient to minimize convex functionals for linear measurements [3-5]. Its application to nonlinear measurements enjoys further benefits, such as improving global convergence properties, but it has not been widely explored [6]. In this work, we investigate an iterative regularized scheme based on the Bregman distance to solve the material decomposition problem in spectral CT for one projection image.1. K Taguchi et al, Med Phys, 40(10):100901, 20132. N Ducros et al, Proc. of CT Meeting, 20163. M Burger, Level Set and PDE Based Reconstruction Methods in Imaging, 1-70, 20134. T Goldstein and S Osher, SIAM J Imaging Sci, 2:323–343, 20095. JFPJ Abascal et al, Med Phys, 38(11):6275-84, 20116. M Bachmayr and M Burger, Inv Prob, 25:105004, 2009