The effect of different xsN labeled sources on the estimation of N2 fixation by soybean was investigated in the Philippines, using a non-nodulating isoline as a reference plant. The soil was placed in 1.5 • 1.5 m concrete plots. The labeled urea or rice straw was applied 1 month before soybean was sown in the soil which had received a labeled ammonium salt prior to the cultivation of two crops. Plots without these additions were also used. The N2 fixation during 0-36 days after sowing was negligible. The estimation of N2 fixed at 67 days of plant growth in the plots which received labeled urea or straw was not statistically different from the estimation in the plots with residual ~SN. The estimations based on both the N balance and ~SN dilution methods were not statistically different. The percentage of N derived from N2 was 41 and the quantity of N~ fixed was estimated at 33 kg N/ha. The least variation was found for the estimation of N2 fixation in the plots with residual ~SN. This fact is attributed to the stability of tSN in the soil, as the 15N abundance in soybeans in the plots with residual ~SN only showed the least variation among the replicates and did not change with the sampling time.