AbstractPlug and Werner (Nature 2002, vol. 417, pp. 929–933) have constructed a model for the growth of ice‐wedge polygons. Contrary to field evidence, the model describes development of ice wedges in frozen ground beneath a centre line of ridges and assumes that effects associated with perturbations in the thermal stress field due to growth of individual wedges may dissipate in the long run. In the field, troughs, not ridges, overlie wedges, and development of the troughs as the ice wedges grow increases snow accumulation above the wedges, so that older wedges may crack less frequently than younger ice wedges. As a result, the ‘single‐ridge’ model does not replicate the inexorable evolution of polygonal networks in the field, because the underlying assumptions are inconsistent with field conditions. Instead, the ‘single‐ridge’ ice‐wedge model attributes network development only to periods of high thermal stress, i.e. particularly cold winters. The model has been used to simulate development of such networks ab initio, by considering conditions in a recently drained thaw lake. However, the results do not reproduce field conditions because the model operates in frozen, not freezing ground, and therefore the initial modelled network is greatly exaggerated from known field observations. While the conclusions of the modelling are presented in general terms, the model only considers epigenetic polygons, which comprise a small fraction of the ice‐wedge polygons in polar terrain. Other, more extensive, syngenetic and anti‐syngenetic (hillslope) polygon networks are not mentioned. Copyright © 2004 John Wiley & Sons, Ltd.