The formation of economic‐grade noble metal and heavy mineral placers at high latitudes is facilitated by the interaction of high‐energy waves and currents with the multimodal size and density fractions of glacial material (Emory‐Moore & Solomon, 1989). Placers which form under such conditions tend to exhibit very irregular distribution patterns, such that these deposits are typically not distinguished by traditional sediment dispersal modeling techniques. In an effort to model sediment dispersal patterns in high‐latitude repons more effectively, multivariate geostatistical analyses, which have proven successful in modeling complex sediment dispersal patterns in a number of marine and lacustrine areas, were carried out on sediment geochemical data for 184 samples from a typical high‐latitude coastal site (North Cardigan Bay). A Q‐mode factor analysis of North Cardigan Bay sediments revealed that > 99% of the total variance in the data can be explained by four end members, representing (1) mixed quartz and orthoclase, (2) mixed aluminosilicates, (3) carbonates, and (4) heavy minerals. Linear programming was used to estimate the relative amount of these end members in each sample; the data from this analysis were used to model sediment dispersal. By comparison with the results of a comprehensive study of North Cardigan sediments by Moore (1968), this modeling approach accurately predicts the sediment distribution pattern for North Cardigan Bay. This sediment dispersal modeling procedure represents a significant improvement over traditional techniques, in that sediment dispersal is not constrained to occur along concentration gradients and it requires only the type of sediment geochemical information that would be obtained in a standard reconnaissance‐level exploration survey.