The interest in functional foods, probiotics and prebiotics requires a proper method to determine specific bacterial groups in the intestinal flora, especially bifidobacteria and lactobacilli. Three media for lactobacilli (MRS, Rogosa, LAMVAB), three media for bifidobacteria (RB, NPNL, Beerens medium) and nine media for total anaerobes have been tested for selectivity and recovery. For total anaerobes Faecal Reinforced Clostridial Agar (FRCA) showed the highest cfu/g, followed by Columbia Blood Agar and BHI Blood agar. Then were no significant differences between the media tested. Reduced physiological salt solution was found to be the best dilution medium. For bifidobacteria and lactobacilli samples of human faeces. cat faeces and pig ileal contents were used. Bifidobacteria could reliably be determined on all three media tested in human faeces, but not on pig ileal contents or cat faeces. Absolute counts were highest in human samples. No lactobacilli could be isolated on MRS in either sample, none of the colonies in the countable plates were lactobacilli. For Rogosa over 90% of the colony types observed in human samples were not lactobacilli. For cat faeces this was 58%, but no false positives were observed in the pig ileal samples. For LAMVAB the percentages of false positive colony types were 9, 14 and 0% for human, cat and pig samples. It can be concluded that for bifidobacteria RE and Beerens medium show comparable results, and can be used to quantify bifidobacteria in human faeces, but none of the media tested is suitable for reliably counting bifidobacteria from pig and cat samples. For lactobacilli LAMVAB shows the highest sensitivity. (C) 1999 Elsevier Science B.V. All rights reserved.