A three year trial (2004-2006) was conducted to assess the performance of three forage legumes, Stylosanthes hamata(L) Taub. cv Verano, Chamaecrista rotundifolia (Pers) Grenne cv Wynn and Centrosema pascuorum Mart. ex Benth (Centurion) cv Cavalcade in range rehabilitation. The first phase of the trial covered a period of sixteen months when the feasibility of range improvement by strip-sowing of the selected forage legumes into natural pastures was tested. Data on legume establishment, forage dry matter yield were collected from the strip-sown legumes and the natural pastures. The first phase was followed by a second one where Bunaji bulls were introduced to graze the treatments continuously over the next 12 months. The grazing animals were weighed fortnightly to monitor liveweight changes while the body condition score were recorded on a monthly basis. In the final year, persistency and spread of the different forage legumes were assessed. In the year of establishment, S. hamata recorded the highest stand count of 117 stands/m2, with lower counts of 65 stands/m2 for C. pascuorum and 25 stands/m2 for C. rotundifolia. The S. hamata stand count was significantly higher (P<0.01) than that of either C. pascuorum or C. rotundifolia. In the second year following establishment, stand count in both S. hamata and C. pascuorum dropped while C. rotundifolia increased to 67 stands/m2. The stand counts in S. hamata and C. pascuorum further decreased in the third year while C. rotundifolia stand count increased. Regeneration from seed from the legumes tested was 100% in C. pascuorum, 99.9% in C. rotundfoilia and 87.6% in S. hamata. In the third year, C. rotundifolia was found to have spread beyond its sown strips. Dry matter yields from the strips of the sown legume at commencement of grazing was highest in S. hamata strips (1.4t DM/ha) than C. rotundifolia (1.1t DM/ ha) and C. pascuorum (0.9t DM/ha). The dry matter yield from the native pasture alone was 3.0 t DM/ha. In the grazing trial, paddocks were stocked at 4 Bunaji bulls /ha. The animals grazed from 9:00 am to 5:00 pm over a one year grazing cycle. Weight gain was highest in animals on S. hamata treatment than those on C. rotundifolia, with the least gain recorded in animals on C. pascuorum treatment. However, there was no significant difference due to treatment. Similarly body condition scores, which ranged between 3.30�0.09 to 3.53 � 0.09 were not significantly different (P>0.05) in the treatments. The results of this study, therefore, indicate that C. rotundifolia has the potential for use in re-seeding degraded rangelands. It has the capacity to colonize areas faster than either S. hamata or C. pascuorum. C. pascuorum has the disadvantage of being very susceptible to attacks by various pests which prevented it from setting seed and subsequent regeneration in the open strip-sown range site. It is recommended that Chamaecrista rotundifolia be further tested in the rehabilitation of the degraded rangelands in the other Savanna ecological zones.