A premise of stream restoration theory and practice is that it is often futile to attempt to restore a stream at the reach scale (101-103 metres) until catchment scale problems have been addressed. This study considers reach scale restoration actions undertaken in Bryan Creek, a sand bed river in south east Australia impacted by a sediment pulse, after catchment sediment sources have been addressed. Local scale interventions, which were in-stream sand extraction, fencing to exclude stock and riparian revegetation, were evaluated by quantifying cross-section and thalweg variability, mapping in-stream and riparian vegetation and by classifying the morphology that emerged following each intervention. Following intervention channel reaches moved to one of three distinct states: simple clay bed, eroding reaches dominated by Juncus acutus, and reaches with deep pools and Phragmites australis. Boundaries between the intervention reaches were sharp, suggesting local scale interventions dominate over catchment scale processes. The magnitude and spread of variability metrics were similar between all reaches and differences in variability bore no relation to intervention type, despite the stark difference in post-intervention morphology. These findings suggest that cross-section and thalweg variability metrics are an inadequate proxy for the effectiveness of local scale interventions in accelerating the recovery of sand bed reaches from a bedload pulse. The most important implications for river managers is that local scale interventions can lead to substantial and rapid improvements in condition, and the change in condition of these reaches is almost independent of other reaches. In this case, the key to the pattern of reach scale geomorphic recovery is excluding stock from waterways so that a specific macrophyte can establish, trap sediment and develop pools.