1. This study highlights the use of waterbird communities as potential measures of river and floodplain health at a landscape scale.
2. The abundance and diversity of a waterbird community (54 species) was measured over 15 trips with four aerial and three ground counts per trip on a 300-ha lake in arid Australia.
3. Aerial survey estimates of individual species were significantly less precise (SE/mean) than ground counts across two (11–100 and > 1000) out of four abundance classes of waterbirds: 0–10, 11–100, 101–1000 and > 1000. Standard error/mean as a percentage decreased with increasing abundance from about 60% for the lowest abundance class to 18% for the largest abundance class.
4. Aerial survey estimates were negatively biased for species in numbers of less than 10 and greater than 5000 but unbiased compared to ground counts for other abundance classes. Aerial surveys underestimated numbers of waterbirds by 50% when there were 40 000 waterbirds. Three ground counts found about seven more waterbird species than four aerial surveys. One ground count took about 150 times longer than two aerial surveys and cost 14 times more.
5. Regression models were derived, comparing aerial survey estimates to ground counts for 31 of 36 species for which there were sufficient data. Aerial survey estimates were unbiased for most of these species (67%), negatively biased for six species and positively biased for one species. Estimates were negatively biased in species that occurred in small numbers or that dived in response to the aircraft.
6. River system health encompasses the state of floodplain wetlands. Waterbirds on an entire wetland or floodplain may be estimated by aerial survey of waterbirds; this is a coarse but effective measure of waterbird abundance. Aerial survey is considerably less costly than ground survey and potentially provides a method for measuring river and floodplain health over long periods of time at the same scale as river management.