Continental basalts ranging in age from 16.5 to 19.08 Ma crop out throughout the northern part of the Arabian plate. The basalts have distinctive petrographic characteristics such as rounded and skeletal olivine phenocrysts with abundant melt inclusions, implying the mixing of two distinct magmas. All of the analysed basalts are tholeiitic in composition. The presence of quartz xenocrysts with clinopyroxene rims in some samples indicates that crustal assimilation was probably an important process during magma ascent to the surface, and low Mg number and high SiO2 contents of the basalts clearly show that they have experienced fractional crystallization as well as crustal contamination. Variations of the major and trace elements versus MgO show that olivine + clinopyroxene + plagioclase were the main fractionating minerals. In terms of incompatible trace elements, the basalts have OIB-like signatures with a slight depletion at Nb–Ta on primitive-mantle-normalized diagrams. The basalts have slightly LREE enriched patterns with La/YbN = 5.5 to 6.7. La/Nb ratios are close to unity, suggesting the melts may have originated in the asthenospheric mantle. Partial melting modelling based on REE data imply that the melts were not produced from a single mantle source depth, which is either purely a spinel- or garnet-peridotite end member. The samples lie on a binary mixing line between low-degree melts (< 5 %) from garnet-peridotite and higher-degree melts (> 10 %) from spinel-peridotite sources on a plot of La/Yb v. Dy/Yb, requiring interaction of melts derived from both garnet- and spinel-peridotite fields. Melts originating from both sources were initially tapped by distinct magma chambers, which subsequently hybridized into a single flow. Hybridized magma ascended to the surface along Neogene strike-slip faults, which are linked to the Dead Sea Fault Zone.