Climate change is expected to make dryland rainfall more variable, but it is unclear whether drylands soils respond primarily to historical precipitation pattern a site has experienced or to contemporary changes in water availability. In this study, we assessed the relative importance of aridity of the site, long-term precipitation legacy, and contemporary watering in shaping the rhizosphere, including soil properties, microbial community composition and function (extracellular enzyme activity), and root traits. Soils from a long-term (>5 years) rainfall manipulation experiment (Increased/Decreased, ±65%) at an arid (Broken Hill) and a semi-arid (Nyngan) site in south-eastern Australia were used to grow a common native grass, Dichanthium sericeum under common glasshouse conditions. Plants were eatablished under well-watered conditions and then exposed to watering treatments (±65%) that either matched or mismatched the soil’s precipitation legacy. Our results showed that the response across all measured variables was hierarchical and persisted under common conditions. Site (climate history) was the dominant driver, precipitation legacy was a persistent secondary filter, and contemporary watering had narrower and legacy-dependent effects. Shifts in microbial community composition was similar across both sites. Broken Hill soils showing altered soil chemistry and bacterial diversity, narrow enzymatic response (2), whereas in Nyngan soils, it was expressed through broader and coordinated shifts (4) in enzymatic activity. Cross-compartment correlations showed that rhizosphere components responded as a coupled system. The strength of coupling varied by site and precipitation legacy, with highest cross compartment correlations in soils with decreased precipitation legacy in Broken Hill . Contemporary watering treatment did not override the site or precipitation legacy effects but acted as a legacy- and compartment-specific modifier with a significant effect on root traits and select enzymatic activity, particularly when contemporary watering was mismatched with the precipitation legacy. These results show that predictions of dryland responses to future changes should account for precipitation legacies and rhizosphere coupling and not only the current watering availability.