Susceptible rat strains develop hepatobiliary injury following the surgical creation of self-filling blind loops that cause small bowel bacterial overgrowth. Luminal bacteria or their cell wall polymers were implicated in the pathogenesis of the lesions because sham-operated rats and rats with self-emptying blind loops, having only slightly increased bacterial counts, did not develop hepatic injury. In this study, antibiotics with different spectra of activities were continuously administered starting 1 day or 22 days after surgery to determine which intestinal flora may be responsible for the development of hepatic injury in rats with small bowel bacterial overgrowth. Four weeks following surgery, Lewis rats with self-filling blind loops receiving no antibiotics had elevated liver histology scores (8.2 +/- 1.3 vs. 0.7 +/- 0.4) and plasma aspartate aminotransferase levels (269 +/- 171 vs. 84 +/- 24) compared with sham-operated rats, P less than 0.001. Oral gentamicin as well as oral and intraperitoneal polymyxin B, which binds endotoxin, did not prevent hepatic injury in rats with self-filling blind loops. However, oral metronidazole and tetracycline therapy continuously administered beginning 1 day after surgery diminished hepatic injury (histology score 3.0 +/- 1.8, 2.9 +/- 1.1; aspartate aminotransferase 87 +/- 25, 98 +/- 34; respectively P less than 0.001 compared with self-filling blind loops receiving no antibiotics). Metronidazole also protected Wistar rats that require 12 weeks to develop hepatic injury following experimentally induced small bowel bacterial overgrowth compared with rats with self-filling blind loops that received no antibiotic treatment (histology score 10.4 +/- 1.3 vs. 0.7 +/- 1.1, and aspartate aminotransferase 273 +/- 239 vs. 76 +/- 20, P less than 0.001). When rats started metronidazole therapy 22 days after self-filling blind loop surgery, elevated aspartate aminotransferase values decreased to normal during the next 77 days and final histology scores were normal. All rats with self-filling blind loops had negative peritoneal, liver, spleen, and blood cultures but approximately 75% of mesenteric lymph node cultures were positive irrespective of antibiotic treatment. Because Bacteroides species have been implicated in causing vitamin B12 and disaccharidase deficiencies in rats with self-filling blind loops, we documented the presence or absence of these organisms from blind loops using selective culture techniques. Metronidazole and tetracycline eliminated Bacteroides sp. from blind loops, but polymyxin B and gentamicin did not.(ABSTRACT TRUNCATED AT 400 WORDS)