Global warming has induced more frequent and intense heavy rainfall, significantly affecting dryland productivity, which dominates the interannual trend and variability of the terrestrial global carbon cycle. However, the lasting impact of such extreme wet years, termed as ’legacy effects,’ is poorly understood. We used satellite NDVI and GPP derived from both the LUE model and eddy covariance as productivity proxies to examine legacy effects on productivity after extreme wet years in Australia, where 75% of the continent is drylands, and the climate is highly variable due to ENSO events. After excluding random responses using Superposed Epoch Analysis (SEA), we confirmed that only the 2010-2011 wet year, caused by the La Niña event, indicated a strong and significant legacy effects in the subsequent first year (over 50% of the area showed significant positive legacy effects (> 90% significance threshold) in the first year following). We then conducted a further analysis of the legacy effect and found that the arid areas in Australia show significant sensitivity (p<0.05) to the productivity of the previous year. In these antecedent-sensitive areas, the legacy effects can contribute up to 25% of the current year’s productivity. Furthermore, the productivity of wet years is the main factor (r > 0.9, p < 0.01) driving the legacy effects, rather than the severity of the wet conditions, i.e., it is the plant growth during wet years that caused legacy effect in following year rather than the extreme rainfall itself. Our results highlight that the wet year’s legacy effect should be considered in carbon-climate models in drylands. It is crucial to revisit the legacy effects of wet years on dryland productivity from an ecogeographic perspective: investigating the coupling of abnormal increases in productivity during wet years and their occurrence in antecedent-sensitive regions, as these ecogeographic characteristics are key determinants of the magnitude of legacy effects.