This paper presents a solution for the inverse problem, or interpretation, of the radial flow of tracers in naturally fractured reservoirs. Our model considers cubic-matrix-block fracture geometry, and by using the Rosenbrock method for nonlinear regression, we can estimate up to six parameters for this cubic-block geometry. We carefully tested the nonlinear regression against synthetical tracer-concentration responses affected by random noise with the objective of simulating, as closely as possible, step-injection field data. Results were obtained within 95% confidence limits. We also investigate sensitivity of the inverse-tracer-problem solution on the main parameters that describe this flow problem. The main features of the nonlinear regression program used in this study are also discussed. The procedure can be applied to interpret tracer tests in naturally fractured reservoirs, allowing the estimation of fracture and matrix parameters of practical interest (longitudinal-fracture dispersivity, a, matrix porosity, phi(2) fracture half-width, W matrix block size, d, matrix-diffusion coefficient, D-2, and the adsorption constant, k(d)). This work's method offers a practical alternative for tracer-flow-test interpretation.