Monitoring of the ecological impacts of water abstraction from unregulated streams in the state of New South Wales (NSW), Australia, is challenging because water is abstracted by thousands of geographically dispersed users who pump intermittently according to temporally varying needs and the limitations imposed by licences and access rules. Detailed, quantitative monitoring methods are too costly for widespread routine application because of the size of the state (801 000 km 2 ) and the large number of streams affected by abstraction. We therefore tested the possibility of detecting abstraction impacts on aquatic macroinvertebrates with rapid biological assessment (RBA) methods similar to those that are routinely used for biological monitoring of Australian rivers. We sampled 85 sites on unregulated streams in north-eastern NSW during a period of prolonged and recurring drought, 54 of which were designated as reference sites with respect to water abstraction because upstream entitlement for abstraction was less than 1% of their mean annual flow (MAF). The remaining, non-reference sites had an average of 4% of MAF licensed for upstream abstraction (range 1-20%). Sweep and kick samples were collected at each site in two seasons, and invertebrates were picked for 30 min per sample and analysed at genus level. We found a small but statistically significant overall difference in macroinvertebrate assemblages between the reference and non-reference sites, but the amount of upstream entitlement did not affect the degree to which assemblages at individual sites deviated from site-specific reference conditions. We attribute the absence of evident impact mainly to the low proportions of streamflow abstracted, but limitations of RBA methods may also be a factor. We recommend a risk-based approach to future monitoring whereby effort is focussed on those streams where a high proportion of flow is abstracted.