Aims: To investigate the short- and long-term effects of an extract of Sapindus rarak saponins (SE) on the rumen fibrolytic enzyme activity and the major fibrolytic micro-organisms. Methods and Results: Two feeding trials were conducted. In the short-term trial, four fistulated goats were fed a basal diet containing sugar cane tops and wheat pollard (65 : 35, w/w) and were supplemented for 7 days with SE at a level of 0.6 g kg(-1) body weight. Rumen liquor was taken before, during and after SE feeding. In the long-term trial, 28 sheep were fed the same basal diet as the goats and were supplemented for 105 days with 0.24, 0.48 and 0.72 g kg(-1) body mass of the extract. Rumen liquor was taken on days 98 and 100. Protozoal numbers were counted under the microscope. Cell wall degradation was determined by enzyme assays and the major fibrolytic micro-organisms were quantified by dot blot hybridization. Conclusion: Sapindus extract significantly depressed rumen xylanase activity in both trials and carboxymethylcellulase activity in the long-term trial (P < 0.01). Fibrobacter sp. were not affected by the SE in both trials, while ruminococci and the anaerobic fungi showed a short-term response to the application of saponins. Protozoal counts were decreased only in the long-term trial with sheep. Significance and Impact of the Study: These data suggest that there is an adaptation of Ruminococcus albus, Ruminococcus flavefaciens and Chytridiomycetes (fungi) to saponin when fed over a long period. The fact that no correlation between the cell wall degrading enzyme activities and the cell wall degrading micro-organisms was observed suggests that the organisms tracked in this experiment are not the only key players in ruminal cell wall degradation. Sapindus rarak saponins partially defaunate the rumen flora. Their negative effect on cell wall degradation, however, is not related to rumen organisms currently recognized as the major cell wall degrading species. The adaptation of microbes in the long-term feeding experiment suggests that the results from short-term trial on the ruminal microbial community have to be interpreted carefully.