A 5.7-kilobase EcoRI fragment, PG2I, and a 3.9-kilobase EcoRI fragment, PG80I, were cloned randomly from the total DNA of Pseudomonas glumae, the causal agent of bacterial grain rot of rice. PG2I hybridized to all the 43 strains of P. glumae isolated from rice grain, rice seedlings, rice leaf sheaths and mung bean sprouts from different places, but PG80I did not hybridize to 5 of the 43 strains. PG2I seemed specific to this pathogen because this fragment failed to hybridize to selected strains of gram negative bacteria including P. gladioli pv. gladioli, P. gladioli pv. allicola, P. caryohylli, P. plantarii, P. solanacearum, P. cepacia, P. avenae, P. cichorii, P. aeruginosa, P. syringae pv. phaseolicola, P. syringae pv. syringae, P. viridiflava, P. marginalis pv. marginalis, Agrobacterium radiobacter, A. rhizogenes, A. tumefaciens, Erwinia ananas, E. carotovora subsp. carotovora, E. herbicola, X. campestris pv. campestris, and X. campestris pv. oryzae, and gram positive bacteria including Arthrobacter illicis, Clavibacter michiganensis subsp. michiganensis, Cl. michiganensis subsp. sepedonicus, Cl. rathayi, Curtobacterium albidum, C. citreum, C. flaccumfaciens pv. flaccumfaciens, C. flaccumfaciens pv. oortii and Rhodococcus fasciens. When PG2I was separated into two fragments, PG2Ia and PG2Ib, by digestion with KpnI, PG2Ia hybridized to all the strains but PG2Ib did not hybridize to 18 of the 43 strains of P. glumae, indicating that PG2Ia is specific to this pathogen. Hybridization patterns of PG2Ib and PG80I to these strains showed that they were divided into 4 groups and that at least three groups were present in three localities in Oita Prefecture in 1984. These findings suggest allopatric differentiation of different strains of P. glumae. These probes are considered to be useful for the detection of P. glumae in ecologial studies.