Small ponds are often the main source of drinking water for grazing livestock. The hydrology of these ponds must be understood so impoundments can be located, designed, and managed to avoid water shortages during dry weather. A study was conducted to measure the water balance of a stock-watering pond in the Flint Hills region of east-central Kansas from June 2005 to October 2006. The 0.35-ha pond supplied water to 250-kg yearling steers in a 65-ha pasture of native tallgrass prairie. Evaporation, depth change, and cattle consumption were measured continuously using meteorological sensors, depth recorders, and water meters. Seepage, transpiration, and inflow were measured periodically or modeled. Evaporation was also predicted from weather data using forms of the Penman and Priestley-Taylor models. Evaporation accounted for 64% of the total water loss annually, while seepage, cattle consumption, and transpiration accounted for 31%, 3%, and 2%, respectively. The greatest water loss was observed in July, with total monthly losses over 358 mm and peak daily losses sometimes exceeding 18 mm · d⁻¹. Cattle consumption averaged 30 L · day⁻¹ · animal⁻¹ with peak usage of 46 L · day⁻¹ · animal⁻¹. On average, the Priestley-Taylor and Penman evaporation models estimated monthly evaporation to 3% and 5%, respectively. Thus, evaporation, the main form of loss, can be predicted with simple models using data from weather station networks. Inflows from runoff proved difficult to predict and were highly dependent on antecedent soil water content. Results showed that losses from ponds can be measured or predicted with reasonable accuracy. These data could be incorporated into catchment-scale hydrology models to provide site-specific designs for stock-watering ponds and livestock-watering strategies.