We use recent observations of Solar system orbital motions in order to constrain f(T) gravity. In particular, imposing a quadratic f(T) correction to the linear-in-T form, which is a good approximation for every realistic case, we extract the spherical solutions of the theory. Using these spherical solutions to describe the Sun's gravitational field, we use recently determined supplementary advances of planetary perihelia, to infer upper bounds on the allowed f(T) corrections. We find that the maximal allowed divergence of the gravitational potential in f(T) gravity from that in the teleparallel equivalent of General Relativity is of the order of 6.2 × 10−10, in the applicability region of our analysis. This is much smaller than the corresponding (significantly small too) divergence that is predicted from cosmological observations, as expected. Such a tiny allowed divergence from the linear form should be taken into account in f(T) model building.