Grasses such as sugarcane (Saccharum spp.), maize (Zea mays L.) and sorghum (Sorghum bicolor L. Moench) with C4 photosynthetic system are important crops that can host endophytic bacteria capable for biological nitrogen fixation (BNF). However, fixation measurements have yielded conflicting results. We determined BNF of five sugarcane and two sorghum varieties, maize and Guinea (Panicum maximum Jacq), elephant (Pennisetum purpureum Schum) and buffel (Cenchrus ciliaris L.) grasses in a pot experiment, using the 15 N natural abundance technique, with a soil naturally rich in 15 N (δ 15 N > 16 ‰) and two reference species (castor bean, Ricinus communis L., and cotton, Gossypium hirsutum L.) that absorb only N from the soil (average δ 15 N = 9.2 ‰). The initial results showed that no fixation occurred in elephant and buffel grasses. Sorghum, maize, Guinea grass and four sugarcane varieties had δ 15 N signals significantly lower than both reference species, indicating that they absorbed N from the atmosphere (δ 15 N = 0 ‰). The differences in signals from the reference species translates into proportions of plant N content originating from BNF varied from 12.8 to 19.4 % for sugarcane, 22.4% for maize, 20.9% for Guinea grass, and 24.7 to 31.2 % for sorghum. Nitrogen fixation by these species represents a potential of great fertilizer economy and high yields in low input agriculture.