Deep sandy soils cover more than 10 M ha of Australia's southern dryland cropping lands. Farming these soils is inherently challenging, and yield gaps are often wide due to multiple cropping constraints. Physical constraints that limit rooting depth are commonplace and often attributed to traffic-induced soil compaction. However, recent observations draw attention to a reversible hardsetting behaviour in response to soil water content. This hardsetting phenomenon is poorly understood and it may explain inconsistent crop responses to deep ripping in sandy soils. We examined the dynamic behaviour of soil strength under different soil water contents in deep ripped and unripped sands from four South Australian sites. We analyzed the elemental, mineralogical, chemical and physical compositions to differentiate the nature of hardened subsurface layers. Soil compaction and cementation were ruled out due to the soil's bulk density (<1.6 g cm−3) encountered and the rapid slaking that occurred when a hardened pan was submerged in water. The hardsetting behaviour was confirmed by the abrupt change in soil strength in response to the soil water content, measured using intact samples in a laboratory-based penetrometer. This significant increase in soil strength was also observed for soils collected from the ripped plots and for repacked soil samples, with penetration resistance ranging from 0.4 to 5.7 MPa depending on soil content. In contrast to our hypotheses, the hardsetting layer could not be distinguished from the layers above or below it by any analytical method. Given soil strength is typically measured on field moist soil, and standard chemical analysis did not differentiate hardsetting behaviour, alternative methods of diagnosis are needed to understand where deep tillage is likely to have the desired long-term effect. Identifying the nature of potential binding agents responsible for increased strength upon drying would be valuable. Our results confirmed reversible hardsetting in deep sandy soils is driven by small changes in soil water content, from 0.08 g g−1 to 0.05 g g−1 upon drying. Given that rapid re-consolidation of the soil matrix after expensive deep ripping interventions are reported, there is a strong imperative to improve diagnosis, and to identify if soil ameliorants could improve longevity by disrupting hardsetting behaviour.