Aerial surveys can efficiently monitor populations of large animals in geographically extensive management units but are time-and cost-intensive. Biologists often design population surveys to balance cost, spatial coverage, and precision in estimates of population size. Demographic ratio data from aerial surveys are an important component of population monitoring and management, yet the error of ratio data remains under-appreciated, especially for low-density populations. We quantified error in juvenile:female, male:female, and young:mature male ratios from 177 helicopter surveys of white-tailed deer (Odocoileus virginianus) in 37 wildlife management units encompassing 3,269 km 2 of semi-arid grasslands and brush in South Texas, USA, during 2011-2015. We classified error (the % difference between the mean and the limits of the confidence interval) of surveys as appropriate for research-based (≤25% error) or monitoring surveys (≤50% error). Overall, error usually met criteria for monitoring surveys but not for research-based surveys; male:female (x̄= 32% error) and juvenile:female (x̄= 37%) ratios had smaller errors than young:mature male ratios (x̄= 51%). As number of detections increased, errors declined; minimum number of detections for both components of a given ratio needed for monitoring and research-based surveys was about 30-40 and 150-185 at the 90% confidence interval and 45-60 and 210-230 at the 95% confidence interval, respectively. Acceptable error to reliably detect changes in demographic ratios may require greater survey effort or a different survey design for low-density populations. Furthermore, obtaining sufficient