Exploration for economic nickel sulphide mineralisation has often been directed by rationalisations based on poorly understood genetic models. Among those that have enjoyed popularity over the past decade are the "layered gabbro" model and the "basal ultramafic" or Kambalda model. The new discoveries in the Mt Keith-Agnew area could well become a model for future nickel search. Exploration based on even the crudest state- ment of the Kambalda model - prospect the basal contact of Archaean ultramafic complexes for gossanous expressions of sulphide mineralisa- tion - has met with remarkable success. Refine- ments to this model, besides satisfying the geology of any particular deposit, have the major virtue that many of the genetic options - magmatic segregation, hydrothermal transport, sulphurisation, replacement sedimentary, structural control, etc. - are still open. Recent descriptions of highly magnesian silicate melts associated with the ultramafic rocks, do however, place some constraints on the formation of these deposits. Further evidence, from experimental studies on sulphide-silicate and sulphide-oxide phase relationships, and nickel-iron exchange in pyrrhotite, is taken with descriptions of sulphide nickel occurrences in Western Australia to examine genetic models for existing ore bodies.