First leaves of barley line Riso 5678R, with the recessive mlo5 allele conditioning highly efficient, papilla-based penetration resistance to field isolates of Blumeria graminis (DC.) Speer (Syn. Erysiphe graminis DC.) were used. Leaves were subjected to a double inoculation procedure ("inducer" followed by "challenger") using two fungal isolates. Isolate GE3 is a wild-type, unable to infect Riso-R, and is termed "mlo-avirulent". Isolate HL3/5, selected from GE3, is capable of infecting Riso-R, and is termed "mlo-virulent". When HL3/5 was used as inducer, some attacks penetrated successfully and formed haustoria within epidermal cells. When either isolate was inoculated onto the same leaves as challenger 48 h later, attacks on cells containing an inducer haustorium were almost invariably successful. Thus, cells containing an inducer haustorium showed almost complete induced accessibility to challenge attacks, even by the avirulent isolate. Accessibility was also induced to someextent in adjacent cells, but the effect was localized. By contrast, where inducer attacks failed, and a papilla was formed in the attacked cell, later challenge attacks on the same cells always failed. These, and adjacent cells, showed almost complete induced inaccessibility. Induced inaccessibility, therefore, was more effective in preventing penetration than inherent resistance due to mlo. Papillae formed in inaccessible cells were larger than in controls, which might be related to their efficient penetration resistance. Autofluorogens accumulated in papillae and locally in surrounding host cell wall areas. Some evidence suggested that accessibility may be related to suppression of localized autofluorescent host cell responses, and, conversely that inaccessibility may be related to their enhancement.