Newly-opened flowers of several cultivars were inoculated with conidia of 7 isolates. Infection in pistils, stamens, petals, sepals and receptacles was detected by UV microscopy after softening fixed tissues in 1 &smallcap;N NaOH and staining in 0.1% aniline blue. For all isolates, cultivars and environments, conidia germinated on the stigmas and their hyphae grew into the transmitting tissue of the styles, but so slowly that they sometimes took 4-6 weeks to reach the style bases. Hyphae grew from the style into the carpel only in cv. Troubadour but growth there was limited and did not progress into the receptacle. The fungus produced conidiophores and conidia on the stigmas of 15 of the 23 cultivars tested in the field but did not do so in the growth chamber or greenhouse tests. It readily colonised the anthers and connective between the anther lobes and sporulated on anthers in the field; in some cultivars it grew to the base of a few stamen filaments and into the receptacle. In the greenhouse, conidia remained ungerminated on the petals and sepals for at least 6 d, but in a moist chamber they germinated and hyphae rapidly penetrated the epidermis and colonised internal tissues of these flower parts. Results indicated that stamens may be more important than styles as a source of latent infection.