In the UK, conidia of Pseudocercosporella capsellae, the anamorph of Mycosphaerella capsellae, were observed on white leaf spot lesions on leaves throughout the growing season. Ascomata were not observed on lesions on either green or senescent leaves, although stromatic knots and spermogonia were occasionally seen in summer. However, spermogonia and protoascomata were produced in white leaf spot pod and stem lesions in early summer. Protoascomata continued to mature after harvest in these lesions on the debris. Mature ascomata subsequently developed by early autumn, but were exhausted by early January and did not overwinter. A diurnal periodicity in numbers of air-borne M. capsellae ascospores discharged from infected debris was observed with a Burkard spore sampler, with greatest numbers of ascospores collected near the middle of the day; the records also suggested that ascospores were released in response to wetting by dew or rain. Studies of natural white leaf spot epidemics in winter oilseed rape provided evidence that air-borne ascospores are the primary inoculum for initiating epidemics in the autumn in the UK. White leaf spot disease gradients over 100 m across a winter oilseed rape crop at Rothamsted were fitted by both negative exponential and inverse power-law models, with gradient slopes suggesting the deposition of air-borne spores dispersed from a single local source of inoculum. In comparison, no obvious white leaf spot gradients were observed over 250 m in a severely diseased crop near North Petherton, Somerset, suggesting that the air-borne spores were dispersed from a number of more distant sources in the area. Both patterns of disease were unlikely to have been initiated by P. capsellae conidia, which are dispersed only very short distances by rain-splash. However, once epidemics have been initiated by air-borne ascospores in the autumn, subsequent disease spread within an infected crop is dependent only on splash-dispersed conidia. A revised disease cycle of the pathogen is proposed.