Understanding of the mechanisms involved in the restoration of degraded soils is necessary in order to assess the sustainability of a cropping system. For this purpose, an 8-year experiment was set up to test the efficiency of four cropping systems involving applications of fertilizer and/or manure and incorporation of crop residues and/or fallows on the regeneration of a Haplic Acrisol degraded by about 90 years of traditional cropping (groundnut-millet rotation without any input). The supply of nitrogen by the soil and particle size and chemical distribution were measured and used to characterize the soil organic matter (SOM) after 2 and 8 years of treatments, and then related to nutrients uptake. Two of the four cropping systems tested that were expected to regenerate soil fertility (Q1 and Q2: 4-year rotations including a plowed-in fallow, incorporation of crop residues, and mineral fertilization) both had a depressive effect on SOM content, although the losses of total C and N observed during the experiment (e.g., -23% N and -20% N for Q1 and Q2, respectively) were less than that for the control (traditional cropping B, -33% N). Only the multiyear fallow did not result in further degradation of SOM, without, however, enriching the soil. On the other hand, an annual application of 10 Mg dry matter of manure ha -1, combined with fertilizer, markedly increased the SOM (total N increased by factor 2 after only 2 years), an effect that seemed, however, stabilized at the end of the 8 years. Only a strong (10 Mg dry matter ha-1) and annual supply of manure, associated with mineral fertilization, had a positive effect on SOM, but its durability is not established.