ABSTRACT Using hydrodynamical N-body simulations, we show that the observed structure and kinematics of the extremely metal-poor dwarf irregular galaxy DDO 68 is compatible with a merger event with at least two smaller satellite galaxies. We were able to obtain a self-consistent model that simultaneously reproduces several of its observed features, including the very asymmetric and disturbed shape of the stellar component, the overall ${\rm H\, \small {\rm I}}$ distribution and its velocity field, the arc-like stellar structure to the west, and the low surface brightness stellar stream to the north. The model implies the interaction of the main progenitor of DDO 68 with two systems with dynamical masses $7\times 10^8\, \mathrm{ M}_{⊙ }$ and almost $10^8\, \mathrm{ M}_{⊙ }$ – 1/20 and 1/150 times the dynamical mass of DDO 68, respectively. We show that the merger between DDO 68 and the most massive of its satellites offers a route to explain the large offset of DDO 68 from the mass–metallicity relation. Assuming that the interacting galaxies have metallicities prior to the merger compatible with those of galaxies with similar stellar masses, we provide quantitative evidence that gas mixing alone does not suffice at diluting the gas of the two components; according to our simulations, the ${\rm H\, \small {\rm II}}$ regions observed along the cometary tail trace the low metallicity of the accreted satellite rather than that of DDO 68’s main body. In this case, the mass corresponding to the low metallicity is that of the secondary body and DDO 68 becomes consistent with the mass–metallicity relation.