Eleven cassava genotypes were tested against cassava mosaic disease (CMD) and compared to a local susceptible cultivar in field conditions from June 2011 to July 2012 in Central African Republic (CAR) at two sites representative of the savanna (Damara) and forest (Pissa) zones of the country. The mean number of whiteflies observed on plants varied among genotypes within each site, but was found nearly three times higher at Damara than at Pissa, resulting in a CMD incidence nearly five times higher at Damara than at Pissa. However, no relation was observed between the number of insect on the plants and the level of susceptibility/resistance of the genotypes. The difference of disease pressure between the two sites revealed high level of resistance in several genotypes, while some other ones indicated rather only a partial resistance. Nevertheless, none of the genotypes tested was found immune, in the end, the virus being detected at least in one site in every genotype, including those ones presenting no symptoms in both sites. The impact of CMD on yield components was assessed on the local susceptible check and three partially resistant genotypes, showing that the disease has no significant effect on the tuberous roots number as well as their weight in both sites. The yield potential varied among different genotypes and between the two sites, the mean number of tuberous roots as well as their mean weight being higher in Damara than in Pissa. This study identified highly resistant genotypes such as 'Gabon' that performed well in both sites, and '91/02322' that was symptomless and presented a yield potential equivalent to the local check. These genotypes could be distributed to growers with the main advantage to be resistant to CMD and, therefore, reducing the risk to spread sources of inoculum all over the cassava cropping areas in the country.