Genetic structuring of populations reflects the interaction of genetic drift, mutation, migration and selection, with influences from life history. Aphids are interesting in this regard as they have the potential for unusually high levels of dispersal and natural selection, which typically counter each other. In the present study, winged grain aphids Sitobion avenae (F.) were collected in four 12.2-m high suction traps along a north–south transect in Britain in order to eliminate sampling bias from plant hosts (cereals and grasses; Poaceae), it being known that these insects show host adaptation demonstrable using molecular markers. Samples were analysed at four polymorphic microsatellite loci over two consecutive years. Population allele frequencies were similar nationally during the two years, although clonal diversity varied greatly between sites and years. In the first sampling year following a harsh winter, diversity was found to display a latitudinal clinal trend: the proportion of unique clones (genotypes) increased with latitude. However, this pattern was less apparent the following year, after a milder winter. Nonetheless, overall FST analysis showed that there was little spatial genetic structuring in either sampling year. These data support the view that the insect is highly migratory and also support a theoretical model and previous data suggesting that the reproductive mode is clinal in S. avenae. This appears to be because natural selection (reduced reproductive success of asexual genotypes under cold conditions) is sufficiently powerful to overcome the homogenizing effects of strong migration. There was no clear evidence for isolation by distance for the genetic data obtained. The data are compared with similar data from other aphid species and other insects. Only by the collection of such data sets can an accurate picture be built up relating genetic variability to flight behaviour, including migratory ambit in this group of insects since, due to their small size and rapid dilution in the air, other marking approaches are impracticable over large geographical distances.