The Io plasma torus shows an unanticipated variation of ion temperature as a function of System III (magnetic) longitude λIII. The simple expectation for the perpendicular temperature T⊥, based on ion pick‐up in a tilted dipole magnetic field, is a double‐peaked function with two maxima and two minima per 360° of longitude, having an average value of ∼480 eV for S+ and a relatively small variation amplitude of ∼±7 eV. Instead, the observed T⊥(S+) near Io's orbit has only one maximum and one minimum, with a much smaller average value of ∼30 eV and a much larger relative variation amplitude of ∼±50%. To explain the observed temperature variation, we propose that the longitude variation of the Pedersen conductance of Jupiter's ionosphere causes the pickup speed of ions injected near Io to vary with longitude. We obtain a satisfactory fit to the observed temperature variation if we make the straightforward assumption that the ionospheric conductance varies inversely with the strength of the ionospheric magnetic field. Both the magnitude and phase of the observed T⊥ variation are consistent with this simple model. The observed parallel temperature T∥(S+) has a somewhat smaller average magnitude and a similar longitude variation, but shifted about 70° in phase toward larger λIII. The T∥ behavior is consistent with the idea that T∥ derives mainly from T⊥ through pitch‐angle scattering in the presence of the observed corotation lag of torus plasma. The 70° phase shift requires a scattering timescale of ∼50 hours, consistent with the classical timescale for ion‐ion Coulomb scattering.