Soil compaction is a major contributor to forest soil degradation. To make a decision on soil remediation, both the soil compactness after heavy traffic of forest machinery and the natural soil recovery rate must be estimated. We estimated the impact of heavy traffic on soil specific volume (V) and its recovery rate on two forest sites by yearly collecting steel cylinders close to field capacity, as currently recommended, at different depths (D) during 3 to 4 yr. Though collected at water contents (w) as homogeneous as possible, the comparison of sample volumes led to inconsistent results. Using w as covariate was necessary to quantify the initial compaction and the V recovery with time and D. Moreover, compared to the soil V and w determined with shrinkage analysis, some field values were very large, suggesting an artifact due to hammering the cylinder at large w. The surface layer (0–10 cm) of the less compacted site showed no residual compactness 3 yr after heavy traffic and the 10‐ to 20‐cm layer compactness decreased significantly with time. The compactness of the second site decreased significantly only in the 0‐ to 10‐cm layer, and the recovery was still ongoing after the third year of monitoring. This site had less swelling clays and larger clay content. The recovery of the soil volume was attributed to shrink‐swell processes. Longer monitoring is required to validate these trends, and further research should evaluate the need for more accurate monitoring based on shrinkage analysis and the use of soil organic carbon (SOC) and clay content as covariates.