Multiple geographical regions have been proposed for the domestication of Equus caballus. It has been suggested, based on zooarchaeological and genetic analyses that wild horses from the Iberian Peninsula were involved in the process, and the overrepresentation of mitochondrial D1 cluster in modern Iberian horses supports this suggestion. To test this hypothesis, we analysed mitochondrial DNA from 22 ancient Iberian horse remains belonging to the Neolithic, the Bronze Age and the Middle Ages, against previously published sequences. Only the medieval Iberian sequence appeared in the D1 group. Neolithic and Bronze Age sequences grouped in other clusters, one of which (Lusitano group C) is exclusively represented by modern horses of Iberian origin. Moreover, Bronze Age Iberian sequences displayed the lowest nucleotide diversity values when compared with modern horses, ancient wild horses and other ancient domesticates using nonparametric bootstrapping analyses. We conclude that the excessive clustering of Bronze Age horses in the Lusitano group C, the observed nucleotide diversity and the local continuity from wild Neolithic Iberian to modern Iberian horses, could be explained by the use of local wild mares during an early Iberian domestication or restocking event, whereas the D1 group probably was introduced into Iberia in later historical times.