Soil erosion of productive top soils is an obstacle for achieving an increased food production in a more sustainable way. The three principles of Conservation Agriculture (CA) of no tillage, permanent soil cover and crop rotations, are often seen as a promising solution. This study was undertaken within the framework of the CA2AFRICA project which aims at understanding the physical effects of CA and the reasons of its (non)adoption in Africa. A field level modeling approach was chosen to assess the effect of three types of CA cropping systems on soil loss, compared to a traditional cropping system for the region of Lake Alaotra in central Madagascar, using the Revised Universal Soil Loss Equation (RUSLE). The most accurate method for estimating erosivity R was based on daily effective rainfall data, resulting in a value of 8487 (SI units). For erodibility K, the average of five estimation methods was taken, resulting in a value of 0.038 (SI units). Three slope scenario’s were chosen, with LS values ranging from 0.6 to 4. Together, these factors form a potential erosion of about 484 ton·ha-1yr-1. The crop cover C was divided into a crop component estimated with percentage of canopy cover, and a mulch component estimated with the Mulch Factor. C-values were determined at half month time intervals for four cropping systems: 1) ‘Traditional’, a two year rotation of upland rice and maize with an average C of 0.56; 2) ‘Stylo 1’, a three year rotation including Stylosanthes guianensi at test field yielding an average C of 0.04; ‘Stylo 2’ as Stylo 1, but for situation at farmers’ fields, yielding an average C of 0.14 and 3) ‘Dolichos’, a two year rotation including Dolichos lablab with an average C of 0.13. Support practice values P were set at 0.4 and 0.1 for respectively the traditional and the CA cropping systems. Resulting annual soil loss (ton·ha-1yr-1) was about 87 for ‘Traditional’, 2 for ‘Stylo 1’, 5.5 for ‘Stylo 2’ and 9 ton·ha-1yr-1 for ‘Dolichos. Although validation with a Unit Plot is necessary, the estimated parameters give an indication of the effect of CA on soil loss and allow for future scaling up of soil loss quantification.