This paper deals with the parallel implementation of reconstruction methods in three-dimensional (3D) cone-beam tomography. Indeed, 3D reconstruction of realistic size volumes requires huge computer resources due to the voluminous data to handle, and the long processing times. Parallel computing, on MIMD computers, seems to be a good approach to manage this problem. In this study, we present different solutions for the parallelization of three conventional reconstruction methods, usable on a class of parallel computers. Depending on the method, we consider two approaches referred as a local and a global approach. Performance results obtained on five different MIMD computers are reported. Two 3D images reconstructed from physical data acquired with the “Morphometer” are presented.