We show that observations of high-redshift Lyα emitters (LAEs) have the potential to provide definitive evidence for reionization in the near future. Using 200-Mpc radiative transfer simulations, we calculate the effect that patchy reionization has on the line profile, on the luminosity function, and, most interestingly, on the clustering of emitters for several realistic models of reionization. Reionization increases the measured clustering of emitters, and we show that this enhancement would be essentially impossible to attribute to anything other than reionization. Our results motivate looking for the signature of reionization in existing LAE data. We find that for stellar reionization scenarios the angular correlation function of the 58 LAEs in the Subaru Deep Field z= 6.6 photometric sample is more consistent with a fully ionized universe (mean volume ionized fraction than a universe with at >2σ confidence level. Measurements in the next year on Subaru will increase their z= 6.6 LAE sample by a factor of 5 and tighten these limits. If the clustering signature of reionization is detected in an LAE survey, a comparison with a Lyman-break or an Hα survey in the same field would confirm the reionization hypothesis. We discuss the optimal LAE survey specifications for detecting reionization, with reference to upcoming programmes.