The modifying effects of flavonoids diosmin and hesperidin during the initiation and post-initiation phases of oesophageal carcinogenesis initiated with N-methyl-N-amylnitrosamine (MNAN) were investigated in male Wistar rats. At 7 weeks of age, all animals except those treated each test chemical alone and control groups were given weekly intraperitoneal injections of MNAN (12.5 mg/kg body weight/injection) for 12 weeks to induce oesophageal neoplasms. For examining the modifying effects of 'initiation' treatment of test compounds, groups of animals were fed the diets containing 1000 ppm diosmin and 1000 ppm hesperidin, and the diet containing both compounds (900 ppm diosmin and 100 ppm hesperidin) for 13 weeks, starting 7 days before the MNAN dosing and then switched to the basal diet. For examining the modifying effects of 'post-initiation' treatment of these compounds, the groups given MNAN and a basal diet were switched to the experimental diets containing diosmin, hesperidin or diosmin combined with hesperidin at 1 week after the stop of MNAN injection, and maintained on these diets for 7 weeks. The other groups consisted of rats given test compounds alone or untreated rats. All animals were necropsied at the termination of the study (week 20) to determine the incidences of oesophageal neoplasms and preneoplasms, blood polyamine levels, and cell proliferation activity estimated by 5-bromodeoxyuridine (BrdU)-labelling index and by morphometric analysis of silver-stained nucleolar organizer regions' protein (AgNORs). A number of oesophageal neoplasms developed in rats treated with MNAN alone (75% and 100% incidences of carcinoma and papilloma, respectively). 'Initiation' feeding of diosmin significantly decreased the incidence of squamous cell carcinoma (P < 0.05). Also, 'initiation feeding' of both compounds singly or in combination caused a significant reduction in the multiplicities of oesophageal carcinoma and papilloma (diosmin, 78 and 58% reduction; hesperidin, 70 and 50% reduction; and the combination regimen, 70 and 30% reduction, P < 0.005). 'Post-initiation' feeding slightly decreased the multiplicities of these oesophageal neoplasms. Also, these dietary regimens reduced the multiplicities of preneoplastic lesions (hyperplasia and severe dysplasia; P < 0.05). There were no pathological alterations in rats treated with both compounds alone or the combined regimen alone or those in an untreated control group. Similarly, feeding of these compounds significantly decreased the expression of cell proliferation biomarkers (BrdU-labelling index and AgNORs number) of the non-lesional oesophageal epithelium (P < 0.05). Blood polyamine concentrations were also lowered in rats given the carcinogen and test compounds, both alone and in combination, when compared with those of rats given MNAN alone (P < 0.05). These findings suggest that diosmin and hesperidin supplementation, individually or in combination, is effective in inhibiting the development of oesophageal cancer induced by MNAN when given during the initiation phase, and such inhibition might be related to suppression of increased cell proliferation caused by MNAN in the oesophageal mucosa.