It has been suggested that a full linkage analysis is a prerequisite for confident paternity testing, by using DNA fingerprinting, in natural populations. These fears are based on a confusion between linkage and linkage disequilibrium and a misplaced assumption that linkage between bands will necessarily reduce the effective number of paternal-specific bands. Several methods for detecting linkage without resorting to the analysis of large sibships are considered, for example, by analysing half-sibships, by band-association, and by altering the experimental conditions used. Even if linkage is present, the magnitude of its effects are unlikely to undermine the accuracy of the technique, given the average levels of variability being detected. We conclude that the effects of linkage are only likely to present a problem when sample sizes are very small or when closely related individuals are being tested together.