Cocoa fermentations were performed in wooden boxes under the following four experimental regimens: beans naturally fermented with wild microflora; aseptically prepared beans with no inoculum; and beans inoculated with a defined cocktail containing microorganisms at a suitable concentration either at zero time or by using phased additions at appropriate times. The cocktail used consisted of a yeast, Saccharomyces cerevisiae var. chevalieri, two lactic acid bacterial species, Lactobacillus lactis and Lactobacillus plantarum, and two acetic acid bacterial species, Acetobacter aceti and Gluconobacter oxydans subsp. suboxydans. The parameters measured were cell counts (for yeasts, filamentous fungi, lactic acid bacteria, acetic acid bacteria, and spore formers, including reisolation and identification of all residual cell types), sugar, ethanol, acetic acid, and lactic acid contents (and contents of other organic acids), pH, and temperature. A cut test for bean quality and a sensorial analysis of chocolate made from the beans were also performed. The natural fermentation mimicked exactly the conditions in 800-kg boxes on farms. The aseptic box remained largely free of microflora throughout the study, and no significant biochemical changes occurred. With the zero-time inoculum the fermentation was almost identical to the natural fermentation. The fermentation with the phased-addition inoculum was similar, but many changes in parameters were slower and less pronounced, which led to a slightly poorer end product. The data show that the nearly 50 common species of microorganisms found in natural fermentations can be replaced by a judicious selection and concentration of members of each physiological group. This is the first report of successful use of a defined, mixed starter culture in such a complex fermentation, and it should lead to chocolate of more reliable and better quality.