A gravity inversion routine was used to calculate crustal thickness variations in the Queen Charlotte Basin, western Canada. First, a ''crustal'' Bouguer anomaly was derived using a digital, well-constrained depth-to-basement map and densities determined from well logs and rock samples. This anomaly was subsequently inverted for crustal thickness assuming a single density contrast at depth and an average crustal thickness provided by seismic refraction data. The results, restricted to that portion of the basin for which there are good quality data, show crustal thickness variations of the order of 12 km. Regions of thin crust are located off northern Vancouver Island (~20 km), along the axis of Hecate Strait and in northern Queen Charlotte Sound (~24 km). Thicker crust is associated with Central Queen Charlotte Ridge and northeastern Hecate Strait. Tests show these trends in crustal thickness prevail for all reasonable ranges of input parameters. Calculated residual gravity anomalies contain information useful for mapping the distribution of volcanics interbedded with sediments in subbasins and for refining estimates of depth-to-basement derived from seismic data alone. Our results are, for the most part, in excellent agreement with seismically derived estimates of Moho depth. They show minimum crustal thickness beneath regions of maximum sediment accumulation, supporting an extensional origin for the basin. They indicate that this extension was quite heterogeneous and yield minimum estimates which vary from ~13% in some parts of the basin to as much as 65% in others. The results can be used to discriminate between models on the Tertiary evolution of the basin; the observed trends are not predicted by the lithospheric flexure or simple margin-perpendicular extension models but are consistent with a model of crustal thinning by distributed shear in a transtensional regime. ¿ American Geophysical Union 1995