Context Ammonia (NH3) volatilisation can be a significant nitrogen (N) loss pathway in the grains industry following the surface broadcast application of urea. However, the extent of urea volatilisation from acidic soils and the soil properties that regulate this N loss pathway have not been investigated widely. Aims We conducted a laboratory incubation experiment to measure NH3 volatilisation loss potential following the broadcast application of urea prills (1–2 mm diameter; 50 kg N ha−1) onto moistened acidic and neutral cropping soils, sampled from four long-term cropping research sites. Methods The selected soils varied in pH, clay content, organic carbon, pH buffering capacity (pHBC) and cation exchange capacity. Volatilised NH3 was captured in a phosphoric acid trap after 7, 14 and 21 days and then measured using colorimetric analysis. We compared the measured NH3 losses with predicted NH3 losses derived from an existing empirical NH3 volatilisation prediction model. Key results Of the applied urea-N, 0.9–25% was volatilised. Cumulative NH3 losses were strongly related (R2 = 0.77) with soil pHBC derived from a pedotransfer function. The existing NH3 loss model generally had poor predictive capacity (RMSE = 34%). Conclusions Using clay content as a surrogate variable for pHBC in the predictive model for sandy kaolinitic soils where it is largely a function of organic carbon content can cause poor estimates of NH3 volatilisation loss potential. Implications Grain production on sandy, acidic soils with low pHBC could lead to substantial NH3 volatilisation losses if urea is broadcast.