jats:titleABSTRACT</jats:title> jats:pWe model the 21-cm signal and Lyman-α emitter (LAE) population evolution during the epoch of reionization in order to predict the 21-cm LAE cross-power spectrum. We employ high-dynamic-range simulations of the intergalactic medium to create models that are consistent with constraints from the cosmic microwave background, Lyman-α forest, and LAE population statistics. Using these models we consider the evolution of the cross-power spectrum for a selection of realistic reionization histories and predict the sensitivity of current and upcoming surveys to measuring this signal. We find that the imprint of a delayed end to reionization can be observed by future surveys, and that strong constraints can be placed on the progression of reionization as late as z = 5.7 using a Subaru–SKA survey. We make predictions for the signal-to-noise ratios achievable by combinations of Subaru/PFS (Prime Focus Spectrograph) with the MWA, LOFAR, HERA, and SKA interferometers for an integration time of 1000 h. We find that a Subaru–SKA survey could measure the cross-power spectrum for a late reionization at z = 6.6 with a total signal-to-noise ratio greater than 5, making it possible to constrain both the timing and bubble size at the end of reionization. Furthermore, we find that expanding the current Subaru/PFS survey area and depth by a factor of three would double the total signal-to-noise ratio.</jats:p>