The Kiglapait intrusion on the coast of Labrador is a large Precambrian layered intrusion of probable lopolithic form. It is well exposed and well preserved, probably displaying at the surface all the major rock types generated. The magma was emplaced between the anorthosite of the Nain massif (to the south) and basement complex gneisses and metasediments (to the north), the latter rocks having been deformed plastically so as to conform to the intrusion margin. Younger intermediate intrusives cut the north margin, and younger granite occupies joint planes in the southern portion. The structure defined by layering is that of a closed syncline or elongate basin, with a sharply curved axial line convex to the northeast. Five major zones are recognized: (1) an Outer Border Zone of fine-grained, layered, granular and contaminated melagabbros to ferrodiorites; (2) an Inner Border Zone of subophitic olivine gabbro; (3) a Lower Zone of troctolite; (4) an Upper Zone of olivine gabbro to ferrosyenite; and (5) an Upper Border Zone of fine-grained troctolite to ferrodiorite with inverted stratigraphy. Several subzones can be distinguished on the basis of incoming cumulus minerals. The initial liquid, after giving rise to the rapidly deposited marginal Border Zones, produced plagioclase and olivine simultaneously to form the Lower Zone troctolite, comprising about 80 percent of the intrusion volume. New cumulus phases appearing in the Upper Zone are, in upward stratigraphic order, clinopyroxene, magnetite-ulvöspinel and ilmenite, apatite, antiperthite (in place of plagioclase), pyrrhotite, mesoperthite (in place of antiperthite), and K feldspar (possibly not a cumulus phase). The uppermost differentiate is a ferrosyenite. The compositions of major minerals vary regularly with stratigraphic height: plagioclase, from An 62 to An 6 ; olivine, from Fo 72 to Fo 4 (possibly Fo 0 ); and clinopyroxene from augite to hedenbergite. The presence of unoxidized ulvöspinel and the strong trend of enrichment in ferrous iron signify fractionation under severely reducing conditions. The magma chamber was probably a closed system throughout most of its life. Evidence for crystal settling under the influence of currents is present in the form of layering, lamination, cross-bedding, channel scours, and downdip mineral lineation. Slump structures, autointrusion features, and locally drag-folded layers testify to the presence of a crystal mush. The generation of competent units which overlie incompetent units can be explained by cumulate theory, if the competent units represent slow crystal accumulation and isothermal solidification by the adcumulus (diffusion) process, and the incompetent units represent rapid crystal accumulation and delayed solidification by the orthocumulus (fractionation) process. A modification of Hess’s theory on the origin of layering can account for the Kiglapait layering if nucleation and sorting occurred primarily near the roof. Periodic variation in convection current velocity, needed to produce sorting, results from a number of causes, most notably the foundering of metastable plagioclase rafts which now appear as pods or lenses caught in the layered rocks. The geometry of the intrusion is sufficiently simple to allow computation from cross sections of the relative volumes of various rock types, and, in turn, the contouring of the exposed surface in terms of percent solidified. Volume percent of the intrusion solidified is used for the plotting of mineral variation curves and as a basis for eventual chemical calculations. The summed mineral and modal composition of the intrusion results in a bulk chemical estimate of basaltic character. The parent magma was probably high alumina basalt, with poorly developed tholeiitic affinity. The reason for fractionation to ferrosyenite rather than an oversaturated residuum may lie in part in the scant opportunity for assimilation afforded by an anorthosite roof, but the full reason is not yet clear.