New Sr–Nd–Pb–Hf data require the existence of at least four mantle components in the genesis of basalts from the the North Atlantic Igneous Province (NAIP): (1) one (or more likely a small range of) enriched component(s) within the Iceland plume, (2) a depleted component within the Iceland plume (distinct from the shallow N-MORB source), (3) a depleted sheath surrounding the plume and (4) shallow N-MORB source mantle. These components have been available since the major phase of igneous activity associated with plume head impact during Paleogene times. In Hf–Nd isotope space, samples from Iceland, DSDP Leg 49 (Sites 407, 408 and 409), ODP Legs 152 and 163 (southeast Greenland margin), the Reykjanes Ridge, Kolbeinsey Ridge and DSDP Leg 38 (Site 348) define fields that are oblique to the main ocean island basalt array and extend toward a component with higher 176Hf/ 177Hf than the N-MORB source available prior to arrival of the plume, as indicated by the compositions of Cretaceous basalts from Goban Spur (∼95 Ma). Aside from Goban Spur, only basalts from Hatton Bank on the oceanward side of the Rockall Plateau (DSDP Leg 81) lie consistently within the field of N-MORB, which indicates that the compositional influence of the plume did not reach this far south and east ∼55 Ma ago. Thus, Hf–Nd isotope systematics are consistent with previous studies which indicate that shallow MORB-source mantle does not represent the depleted component within the Iceland plume [Thirlwall, J. Geol. Soc. London 152 (1995) 991–996; Hards et al., J. Geol. Soc. London 152 (1995) 1003–1009; Fitton et al., Earth Planet. Sci. Lett. 153 (1997) 197–208]. They also indicate that the depleted component is a long-lived and intrinsic feature of the Iceland plume, generated during an ancient melting event in which a mineral (such as garnet) with a high Lu/Hf was a residual phase. Collectively, these data suggest a model for the Iceland plume in which a heterogeneous core, derived from the lower mantle, consists of ‘enriched’ streaks or blobs dispersed in a more depleted matrix. A distinguishing feature of both the enriched and depleted components is high Nb/Y for a given Zr/Y (i.e. positive ΔNb), but the enriched component has higher Sr and Pb isotope ratios, combined with lower εNd and εHf. This heterogeneous core is surrounded by a sheath of depleted material, similar to the depleted component of the Iceland plume in its εNd and εHf, but with lower 87Sr/ 86Sr, 208Pb/ 204Pb and negative ΔNb; this material was probably entrained from near the 670 km discontinuity when the plume stalled at the boundary between the upper and lower mantle. The plume sheath displaced more normal MORB asthenosphere (distinguished by its lower εHf for a given εNd or Zr/Nb ratio), which existed in the North Atlantic prior to plume impact. Preliminary data on MORBs from near the Azores plume suggest that much of the North Atlantic may be ‘polluted’ not only by enriched plume material but also by depleted material similar to the Iceland plume sheath. If this hypothesis is correct, it may provide a general explanation for some of the compositional diversity and variations in inferred depth of melting [Klein and Langmuir, J. Geophys. Res. 92 (1987) 8089–8115] along the MAR in the North Atlantic.