[en] We analyse two ASCA observations of the highly eccentric O9 III+B1 III binary Iota Orionis obtained at periastron and apastron. Based on the assumption of a strong colliding-wind shock between the stellar components, we expected to see significant variation in the X-ray emission between these phases. The observations proved otherwise: the X-ray luminosities and spectral distributions were remarkably similar. The only noteworthy feature in the X-ray data was the hint of a proximity effect during periastron passage. Although this `flare' is of relatively low significance, it is supported by the notable proximity effects seen in the optical, and the phasing of the X-ray and optical events is in very good agreement. However, other interpretations are also possible. In view of the degradation of the SIS instrument and source contamination in the GIS data we discuss the accuracy of these results, and also analyse archival ROSAT observations. We investigate why we do not see a clear colliding-wind signature. A simple model shows that the wind attenuation to the expected position of the shock apex is negligible throughout the orbit, which poses the puzzling question of why the expected 1/D variation (i.e. a factor of 7.5) in the intrinsic luminosity is not seen in the data. Two scenarios are proposed: either the colliding-wind emission is unexpectedly weak such that intrinsic shocks in the winds dominate the emission, or, alternatively, the emission observed is colliding-wind emission but in a more complex form than we would naively expect. Complex hydrodynamical models are then analysed. Despite strongly phase-variable emission from the models, both are consistent with the observations. We find that if the mass-loss rates of the stars are low then intrinsic wind shocks could dominate the emission. However, when we assume higher mass-loss rates of the stars, we find that the observed emission could also be consistent with a purely colliding-wind origin. A summary of the strengths and weaknesses of each interpretation is presented. To distinguish between the different models X-ray observations with improved phase coverage will be necessary.