In chameleon f (R)-gravity, the fifth force will lead to 'upturns' in galaxy rotation curves near the screening radius. The location of the upturn depends on the cosmic background value of the scalar fieldf R0 , as well as the mass, size and environment of the galaxy. We search for this signature of modified gravity in the SPARC sample of measured rotation curves, using an MCMC technique to derive constraints onf R0 . Assuming NFW dark matter haloes and withf R0 freely varying for each galaxy, most galaxies prefer f (R) gravity to ΛCDM, but there is a large spread of inferredf R0 values, inconsistent with a single global value. Requiring instead a consistentf R0 value for the whole sample, models with log 10 |f R0 | > −6.1 are excluded. On the other hand, models in the range −7.5 < log 10 |f R0 | < −6.5 seem to be favoured with respect to ΛCDM, with a significant peak at -7. However, this signal is largely a result of galaxies for which the f (R) signal is degenerate with the core/cusp problem, and when the NFW profile is replaced with a cored halo profile, ΛCDM gives better fits than any given f (R) model. Thus, we find no convincing evidence of f (R) gravity down to the level of |f R0 | ∼ 6 × 10 −8 , with the caveat that if cored halo density profiles cannot ultimately be explained within ΛCDM, a screened modified gravity theory could possibly provide an alternative solution for the core/cusp problem. However, the f (R) models studied here fall short of achieving this.