Streptococcus gordonii produces two a-amylase-binding proteins, AbpA and AbpB, that have been extensively studied in vitro. Little is known, however, about their significance in oral colonization and cariogenicity (virulence). To clarify these issues, weanling specific pathogen-free Osborne-Mendel rats, TAN : SPFOM(OM)BR, were inoculated either with wild-type strains FAS4-S or Challis-S or with strains having isogenic mutations of abpA, abpB, or both, to compare their colonization abilities and persistence on the teeth. Experiments were done with rats fed a sucrose-rich diet containing low amounts of starch or containing only starch. The mutants and wild-types were quantified in vivo and carious lesions were scored. In 11 experiments, S. gordonii was a prolific colonizer of the teeth when rats were fed the sucrose (with low starch)-supplemented diet, often dominating the flora. Sucrose-fed rats had several-fold higher recoveries of inoculants than those eating the sucrose-free, starch-supplemented diet, regardless of inoculant type. The strain defective in AbpB could not colonize teeth of starch-only-eating rats, but could colonize rats if sucrose was added to the diet. Strains defective in AbpA surprisingly colonized better than their wild-types. A double mutant deficient in both AbpA and AbpB (abpA/abpB) colonized like its wild-type. Wild-types FAS4-S and Challis-S had no more than marginal cariogenicity. Notably, in the absence of AbpA, cariogenicity was slightly augmented. Both the rescue of colonization by the AbpB " mutant and the augmentation of colonization by AbpA " mutant in the presence of dietary sucrose suggested additional amylase-binding protein interactions relevant to colonization. Glucosyltransferase activity was greater in mutants defective in abpA and modestly increased in the abpB mutant. It was concluded that AbpB is required for colonization of teeth of starch-eating rats and its deletion is partially masked if rats eat a sucrose-starch diet. AbpA appears to inhibit colonization of the plaque biofilm in vivo. This unexpected effect in vivo may be associated with interaction of AbpA with glucosyltransferase or with other colonization factors of these cells. These data illustrate that the complex nature of the oral environment may not be adequately modelled by in vitro systems.