DNAs from both prokaryotic and eukaryotic organisms have yielded restriction fragments which manifest markedly anomalous electrophoretic behaviour (reduced mobility) when run on polyacrylamide gels. We have shown previously that the abnormal electrophoretic behaviour of one such fragment is a consequence of stable curvature of the helix axis in solution. The molecules involved tend to contain oligo(dA)-oligo(dT) runs which are approximately in-phase with the helix repeat; however, the precise structural elements responsible for DNA curvature have not been identified. One popular model for curvature invokes a non-coplanar 'wedge-like' conformation of ApA/TpT dinucleotide pairs. Despite a lack of direct evidence in support of this model, it has been used to provide quantitative estimates of curvature. To critically evaluate the ApA wedge model, we have performed an electrophoretic analysis of a series of closely related DNA polymers in which oligo(dA)-oligo(dT) runs of different polarity were compared. We conclude that ApA dinucleotide wedges cannot account for DNA curvature. Therefore, quantitative estimates for ApA wedge deformations, based solely on apparent curvature, cannot be correct.