We study the capture and subsequent annihilation of inelastic dark matter (DM) in the Sun, placing constraints on the DM-nucleon scattering cross section from the null result of the IceCube neutrino telescope. We then compare such constraints with exclusion limits on the same cross section that we derive from XENON1T, PICO and CRESST results. We calculate the cross section for inelastic DM-nucleon scattering within an extension of the effective theory of DM-nucleon interactions which applies to the case of inelastic DM models characterised by a mass splitting between the incoming and outgoing DM particle. We find that for values of the mass splitting parameter larger than about 200 keV, neutrino telescopes place limits on the DM-nucleon scattering cross section which are stronger than the ones from current DM direct detection experiments. The exact mass splitting value for which this occurs depends on whether DM thermalises in the Sun or not. This result applies to all DM-nucleon interactions that generate DM-nucleus scattering cross sections which are independent of the nuclear spin, including the "canonical" spin-independent interaction. We explicitly perform our calculations for a DM candidate with mass of 1 TeV, but our conclusions qualitatively also apply to different masses. Furthermore, we find that exclusion limits from IceCube on the coupling constants of this family of spin-independent interactions are more stringent than the ones from a (hypothetical) reanalysis of XENON1T data based on an extended signal region in nuclear recoil energy. Our results should be taken into account in global analyses of inelastic DM models.