Isothermal crystallization in bulk has been compared dilatometrically with linear polyethylene samples which have been molded in different ways, including melt orientation and sintering of precipitated powders from dilute solutions. These experiments show significant differences in the crystallization kinetics, which in some cases cannot be removed even by prolonged heating of the melt prior to crystallization. The data have been analyzed by an empirical expression in terms of nucleation and growth processes, according to the theoretical approach of Avrami. The results suggest that, in general, the melted polyethylene is not in a true thermodynamic equilibrium and includes some ``quasi‐indestructible'' clusters, which act as heterogeneous nuclei in the crystallization process. The nature and the amount of these nuclei depend on the molding, melting, and mechanical history of the sample, so that the semicrystalline material may have a wide variety of structure.