This dataset contains urine frequency and volume data measured from tri-axial accelerometers on Welsh mountain ewes free-grazing two contrasting upland field sites (semi-improved and unimproved pasture) in North Wales, across two seasons each (spring and autumn). The data, were collected using tri-axial accelerometers glued to the hind of Welsh Mountain ewes to study the urination behaviour of free-grazing sheep. Using a Boolean algorithm, the characteristic squatting position that ewes exhibit upon urination was detected in the accelerometer data. Initially the performance of the accelerometers with sheep in urine collection pens was assessed. Data were collected on the volume of each urination event and recorded the time of each observed urination event. This initial data was used to assess whether the accelerometers and Boolean algorithm were successful in identifying urination events, but also to ascertain whether the time spent in the squatting position would correlate with the volume of urine produced (thus allowing the technique to be able to estimate urine volume from squatting time only in subsequent field deployments). Information on when, where and how often livestock urinate are key data to be able to assess the scale and nature of nitrogen pollution arising from grazed agroecosystems. Urine patches deposited by grazing livestock are large sources of emissions of the greenhouse gas, nitrous oxide, due to high concentrations of nitrogen deposited over relatively small areas. These data were collected in the NERC funded Uplands-N2O project (grant award: NE/M015351/1). Full details about this dataset can be found at https://doi.org/10.5285/127afd24-d2cd-457f-b837-2dd5d328f101