As a basic form of pattern analysis, the parameters of dune spacing, defect density, crest orientation and crest length are measured from remote images and treated statistically for dunes at White Sands in New Mexico, the Algodones in California, the Agneitir in Mauritania, and the Namib in Namibia. Statistical populations are identified from frequency plots of dune spacing and crest length, field-scale calculations of defect density, and rose diagrams of crest orientation. Single populations characterize simple dune fields (White Sands), whereas multiple populations characterize compound/complex dunes (Algodones, Namib), and complex dune fields (Agneitir). As time increases, dune fields show an increase in dune spacing and crest length, a decrease in defect density, more tightly clustered crest orientation, and a reduction in the variance associated with measurements of these parameters. The results are consistent with models of dune fields as self-organizing complex systems in which a characteristic pattern emerges as a function of constructional time. Because pattern evolution is a function of time, it may be possible to use pattern analysis to augment current methods of age determination. Statistically defined populations can be used in geomorphic backstripping to unstack generations of simple patterns that give rise to complex patterns, and to reconstruct each generation in terms of construction time and palaeo-wind regime. Copyright © 2006 John Wiley & Sons, Ltd.