The western Junggar region of China is endowed with considerable resources of gold, copper and chromite. Complex geological conditions and diversity deposit types could lead to various geochemical patterns and geochemical anomalies. In this study, compositional balance analysis (CoBA) is demonstrated for recognition of geochemical patterns and mapping geochemical anomalies that are closely associated with gold/copper/chromite mineralization and particular geological units in the western Junggar region. Here, CoBA was based on hierarchical cluster analysis and sequential binary partition technique, which provides a new path for intuitively distinguishing particular relationships between groups of parts that are of interest. To recognize anomalous patterns in stream sediment geochemical data that are closely associated with mineralization, 18 geochemical elements were used to construct 17 balances by means of the CoBA method, of which four key balances were selected for further investigation. Relevant geological information (e.g. mineral deposit occurrences) provides important references for interpreting and validating the results. By comparison of geochemical patterns of the square roots of Au and Cu concentrations with that of factor scores, the results indicate that the CoBA method provides straightforward and robust interpretation of stream sediment geochemical data by suppressing background patterns and enhancing anomalous patterns in the western Junggar, China.