The analysis of WAXS profiles for various polyethylenes indicates that the proper description of a structure needs the introduction of a kind of 'third phase' in addition to the classical crystalline and amorphous phases. The structure of the additional phase is intermediate between that of the amorphous and crystalline phase. With increasing branch content and molecular weight the intermediate phase becomes more similar to the structure of amorphous phase. The experimental evidence for the intermediate phase is derived not only from the crude approximation of WAXS profiles based on the two phase model but also from the unexpected behavior of the parameters of amorphous halo during crystallization. When crystallization is started, an analysis based upon two-phase model results in an apparent increase of the diffraction angle and width of amorphous halo with time above the values anticipated from the range before the start of crystallization. This is caused by the fact that the amorphous fitting function tries to cover a peak of the intermediate component that appears between amorphous halo and (110) reflection of crystalline phase. The conventionally applied two-phase model leads to several serious errors in determination of structural parameters of both phases. The analysis of crystallization kinetics using three-phase model provides additional information on the nature of crystallization itself. (C) 2004 Elsevier Ltd. All rights reserved.