A collaborative study was conducted in 15 laboratories to evaluate 2 different techniques for enumerating Bacillus cereus in foods. A direct plating technique using mannitol-egg yolk-poly-myxin agar and a most probable number (MPN) technique using trypticase-soy-polymyxin broth were compared for the enumeration of high and low populations of B. cereus in mashed potatoes. The collaborative results showed that the overall mean recovery obtained with the low population level was essentially the same by both techniques. However, the overall mean recovery was significantly higher by the direct plating technique at the high population level. A statistical evaluation of the data also showed that the direct plating technique had better repeatability and reproducibility than did the MPN technique at both the high and low population levels. These results suggest that the MPN technique is suitable for examining foods containing low populations of B. cereus, but that the direct plating technique is preferable for foods that contain a high population of this organism. The confirmatory technique used in the proposed method is reliable for presumptive identification of isolates as B. cereus. The method has been adopted as official first action.