Extension estimates based on palinspastic restoration of only the largest faults in a deformed terrane may be significantly deficient due to neglect of "small" faults (i.e., unsampled faults, most of which have smaller displacements than do sampled faults). Small faults individually account for little extension, but there may be large numbers of them. The cumulative extension on small faults can be calculated from the fractal size distributions of fault displacements. Independent extension estimates support the inference from fault population statistics that small faults account for significant amounts of extension. For example, restorations of Mesozoic faults interpreted from seismic reflection profiles across the Viking graben in the North Sea yield extension estimates that are only 40%-75% of independent estimates. The fractal size distribution of fault displacements in the Viking graben is consistent with the hypothesis that the remaining 25%-60% of extension resulted from displacements on small faults.