Plant aboveground biomass (AGB) is a useful metric to assess ecosystem functioning, and its sensitivity to changing environmental conditions provides insight into potential global change impacts. Allometric estimates of AGB using vegetation characteristics such as plant cover or height provide nondestructive biomass proxies for repeated measurements but can introduce uncertainty to estimates. We estimated the relationship between both plant cover and a cover·height index and AGB for 15 plant species from six sites to identify the most reliable approach to estimate biomass nondestructively in semiarid eastern Australian rangelands. Estimates were made by grouping species at four different levels of specificity, to test whether generic estimates were more robust than grouping species based on life history and morphological characteristics. Estimates were then tested on a 1.5-m² plot at each site for validation. In all cases, models were highly significant (P < 0.001) with adjusted R² values ranging from 0.42 to 0.96 for cover models and 0.38 to 0.98 for cover·height index models. We found the addition of height improved model fits in four groups while reducing model fits in two groups. The error around AGB estimates for cover·height index–based models ranged from –66.8 to 4% (absolute mean 35%). Cover-based models had errors between –13.4% and 53% (absolute mean 14.2%). For cover-based estimates of AGB in validation plots, grouping plants by plant functional types (PFTs) increased accuracy (absolute mean error 17.3%) compared with estimates using data from all 15 species (absolute mean of 65.2%). Overall cover was a useful surrogate to estimate AGB (with the exception of one site, accuracy ranged from –2.3% to 11.5%), while height (thought to be a surrogate for canopy characteristics) provided benefit in a few circumstances. We suggest that future research should test additional nondestructive proxies and group species based on PFTs to improve AGB estimates using allometry.