AbstractThe amount of the phytoalexin glyceollin I in root exudate and root hairs of individual seedlings ofGlycine max(L. Merr. cv. Preston) was analysed using a radioimmunoassay.Bradyrhizobium japonicum110spc4, which is able to form nitrogen fixing nodules with this plant, caused an increase of up to 50‐fold in glyceollin I levels in root exudate relative to uninfected control seedlings. Maximum glyceollin I levels were reached within 10 h of incubation. Elevated glyceollin I levels were also observed after incubation of soybean roots in sterile bacterial supernatant, a suspension of autoclaved bacteria or the supernatant from broken cells ofBradyrhizobium japonicum. Increased glyceollin I production is not due to the process of active root hair penetration by the microsymbiont since living bacterial cells are not necessary for the induction. The observed glyceollin I production in response toBradyrhizobium japonicumis several times lower than that after pathogenic infection. Infection with zoospores of the phytopathogenic oomycete,Phytophthora megaspermaf. sp.glycinearace 1, leads within 20 h to an accumulation of 7 nmol glyceollin I/seedling in the root exudate of the compatible cultivar Kenwood and 48 nmol glyceollin I/seedlings in that of the incompatible cultivar Maple Arrow. These results support the idea that phytoalexins are implicated in determination of compatibility in pathogenic interactions. Crude cell extracts of different symbiotic bacteria (Bradyrhizobium japonicum110spc4,Rhizobium meliloti2011,Rhizobium leguminosarumPRE 8,Sinorhizobium frediiHH 103) were found to induce different amounts of glyceollin I in the root exudate. The observed glyceollin I levels could not be correlated with the ability of these rhizobial strains to nodulateGlycine max. Inhibition of flavonoid and phytoalexin synthesis by (R)‐(1‐amino‐2‐phenylethyl)phosphonic acid (APEP), a specific inhibitor of the phenylalanine‐ammonia‐lyase (PAL), during the first 20 h of the symbiotic interaction dramatically decreased the number of nodules formed in root regions that had been in contact with the inhibitor. This effect was observed at concentrations that inhibited neither bacterial nor plant growth. The implications of these findings for the process of nodule initation