Urban stream ecosystems are vulnerable to urbanisation of surrounding land cover and land use. We study 30 sites along two highly urbanised streams in Brisbane, Australia. Fieldwork generated a suite of primary stream health indicators. Geographic information system techniques generated spatially-explicit metrics of land cover and a lumped metric of nearby population that put stress on stream health. Stream health indicators were aggregated into a stream health index, and land-use stress indicators were aggregated into a land-use stress index, using data envelopment analysis (DEA). DEA was then applied to these indices to create an ecological performance index. Dominator analysis generated a set of practical role models for each ecologically underperforming site. A subsequent round of DEA was applied to the stream health index and multiple stress indicators to calculate response elasticities of stream health with respect to specific stress indicators. Empirical findings show widespread deviations beneath best practice, enlightening dominator relationships, and informative variation in response elasticities. Each of these findings can provide guidance to those responsible for allocating scarce resources in an effort to improve the health of Brisbane's urban streams.