Small-angle X-ray scattering data from several experimental and commercial poly(ether urethane) formulations was used to test various scattering models based on different morphologies. The best fits were generally found with 'globular' scattering models based on a distorted one-dimensional lattice or the Percus-Yevick model of liquid structure. Whilst this is not conclusive proof of the morphologies exhibited by the materials studied, these scattering models are roughly consistent with many AFM and TEM studies, which indicated discrete globular or elongated cylindrical microdomains. Moreover, reasons why meandering elongated or finite lamellar microdomains may behave more like globular scattering bodies are discussed. Hence, these models are proposed as a basis for interpreting scattering data from polyurethanes. Other models were unable to fit the observed scattering data adequately. A model based on the weak segregation of copolymers was rejected on the basis of deviations from the observed scattering behaviour in the Porod region. A model based on stacks of ideal (infinite, parallel, flat) lamellae was ruled out, since it was unable to reproduce the observed peak width. The Teubner-Strey model, based on microemulsion structure, was found to be incapable of fitting the data at low q, particularly for strained samples. However, this model appeared to be more successful at reproducing the scattering observed at elevated temperatures. One possible inference is that the morphologies became more akin to microemulsions during heating. (C) 2004 Elsevier Ltd. All rights reserved.