In 1986, two spring barley cultivars were widely grown in the UK for the first time: Klaxon, which carries the powdery mildew resistance alleles Mla7, Mlk and Ml(La), and Natasha (Mla12 + Ml(Ab)). Significant amounts of a third cultivar, Doublet (Mla7+ Ml(La)), were grown for the first time in 1986. The individual resistance genes, and the combination Mla7+Mlk, had previously been used separately in different varieties. Isolates of the mildew pathogen Erysiphe graminis f.sp. hordei, which were virulent on Doublet and Klaxon, were rare up to June 1986. One clone of E. g. f.sp. hordei, virulent on Doublet and Klaxon, increased in frequency from < 1% to 36% from June to October 1986, in samples from the airborne population in Cambridge, UK. This consisted of isolates with apparently identical virulences, responses to fungicides and genetic fingerprints. It also formed the majority of Doublet-virulent mildew sampled from a field of Doublet near Cambridge in 1987. By contrast, isolates virulent on Natasha were already common and genetically diverse in 1985:22 of 100 isolates sampled in October 1985, belonging to 13 races, were virulent. Natasha appeared not to influence the E. g. f.sp. hordei population greatly, as the frequency of Natasha-virulent isolates fell slightly, from 15·5% to 11·7% between June and October 1986. No single clone predominated in the Natasha-virulent population. These results support the view that new epidemics of barley powdery mildew in the UK arise by highly stochastic evolution of E. g. f.sp. hordei populations. They also indicate that varieties with new combinations of previously exposed resistance genes do not necessarily provide durable control of mildew, because the frequency of a virulent clone may rise rapidly.