Human activities have profoundly altered the global nitrogen (N) cycle. Increases in anthropogenic N have had multiple effects on the atmosphere, on terrestrial, freshwater and marine ecosystems, and even on human health. Unfortunately, methodological limitations challenge our ability to directly measure natural N inputs via biological N fixation (BNF)-the largest natural source of new N to ecosystems. This confounds efforts to quantify the extent of anthropogenic perturbation to the N cycle. To address this gap, we used a pair of indirect methods-analytical modeling and N balance-to generate independent estimates of BNF in a presumed hotspot of N fixation, a tropical rain forest site in central Rondônia in the Brazilian Amazon Basin. Our objectives were to attempt to constrain symbiotic N fixation rates in this site using indirect methods, and to assess strengths and weaknesses of this approach by looking for areas of convergence and disagreement between the estimates. This approach yielded two remarkably similar estimates of N fixation. However, when compared to a previously published bottom-up estimate, our analysis indicated much lower N inputs via symbiotic BNF in the Rondônia site than has been suggested for the tropics as a whole. This discrepancy may reflect errors associated with extrapolating bottom-up fluxes from plot-scale measures, those resulting from the indirect analyses, and/or the relatively low abundance of legumes at the Rondônia site. While indirect methods have some limitations, we suggest that until the technological challenges of directly measuring N fixation are overcome, integrated approaches that employ a combination of model-generated and empirically-derived data offer a promising way of constraining N inputs via BNF in natural ecosystems.