The widespread use of heavy equipment during timber harvesting and site preparation can lead to reduced soil productivity and warrants development of new methods to assess compaction. We evaluated the effects of soil particle density, organic matter, particle size distribution, extractable oxides, and plastic and liquid limits on the maximum bulk density (MBD) of forest soils in British Columbia. Soil samples were collected from 33 sites throughout British Columbia, covering the major forest and soil types of the province. The standard Proctor test was used to determine MBD and related parameters, including the gravimetric water content (WMBD) and porosity (fMBD) at which MBD was achieved. The significance levels of single soil properties in predicting MBD were in the order plastic and liquid limits, organic matter, oxalate‐extractable oxides, and particle size distribution. For all samples, liquid limit and clay were most closely related to MBD (R2 = 0.83). Addition of organic matter to the model increased the regression coefficients, and oxidizable organic matter caused a greater increase than did total C. Stratification of the sample set into groups based on plasticity led to higher R2 values in multiple regressions, and different soil properties were important for nonplastic soils than for those with high, moderate, and low plasticity. Prediction with multiple regression explained the most variation in MBD for nonplastic soils, while properties of highly plastic soils explained the least variation in MBD and moderately plastic soils were intermediate. Based on our findings, we propose an approach for using MBD to help better interpret bulk density data in forest soil compaction studies.