Apical membrane antigen-1 (AMA1) is a conserved apicomplexan protein that plays an important but undefined role in host cell invasion. We have studied the fate of Plasmodium falciparum AMA1 (PfAMA1) during erythrocyte invasion by the malaria merozoite, and compared it with that of the Toxoplasma gondii orthologue, TgAMA1. Shedding of the PfAMA1 ectodomain goes essentially to completion during invasion, and occurs predominantly or exclusively via juxtamembrane cleavage at the previously identified sheddase cleavage site, Thr517. Only the resulting juxtamembrane stub of the ectodomain is efficiently carried into the host cell, and this remains distributed around the plasma membrane of the intracellular ring-stage parasite. Inhibition of normal shedding, however, results in proteolysis at an intramembrane, rhomboid-like cleavage site, and PfAMA1 is susceptible to cleavage by Drosophila rhomboid-1, showing that it can be a substrate for intramembrane cleavage but is not normally processed in this manner. In contrast, shedding of TgAMA1 from the surface of extracellular tachyzoites occurs exclusively via cleavage within the luminal half of its transmembrane domain by a rhomboid-like protease. Also unlike PfAMA1, complete TgAMA1 shedding does not accompany Toxoplasma invasion as the intact protein was readily detected on the surface of newly invaded tachyzoites. This work reveals unexpected differences in the manner in which Plasmodium and Toxoplasma shed AMA1 from the surface of invasive zoites, and demonstrates the presence at the malaria merozoite surface of a rhomboid-like protease.