Polyolefins made using most heterogeneous and some homogeneous Ziegler-Natta catalysts show a characteristic broad molecular-weight distribution, generally assumed to be caused by the presence of more than one active site type, each type producing polymer with an instantaneous chain-length distribution that is most probable and having polydispersity of 2. Herein is reported a comparison of two numerical methods for the deconvolution of instantaneous chain-length distributions of linear homo- and copolymers made using multiple-site-type catalysts giving an individual most probable chain-length distribution for each site type, and a methodology is proposed for the efficient deconvolution of measured chain-length distributions.