Measurement of salivary cortisol has been used extensively as a non-invasive alternative to blood sampling to assess adrenal activity in ruminants. However, there is evidence suggesting a considerable delay in the transfer of cortisol from plasma into saliva. Previous studies in cattle have used long sampling intervals making it difficult to characterise the relationship between plasma and salivary cortisol (PLCort and SACort, respectively) concentrations at different time points and determine whether or not such a time lag exist in large ruminants. Therefore, the objective of this study was to characterise the relationship between plasma and salivary cortisol and determine if there is a significant time lag between reaching peak cortisol concentrations in plasma and saliva across a 4.25 h time-period, using short sampling intervals of 10-15 min, following social separation in dairy cattle. Five cows were separated from their calves at 4 days after calving, and six calves were separated from a group of four peers at 8 weeks of age. Following separation, the animals were moved to an unfamiliar surrounding where they could not see their calves or pen mates. The animals were catheterised with indwelling jugular catheters 1 day before sampling. Blood and saliva samples were obtained simultaneously before and after separation. In response to the stressors, PLCort and SACort increased reaching peak concentrations 10 and 20 min after separation, respectively. This suggested a 10 min time lag between peak cortisol concentrations in plasma and saliva, which was further confirmed with a time-series analysis. Considering the 10 min time lag, SACort was strongly correlated with PLCort (P < 0.0001). Salivary cortisol correlates well with plasma cortisol and is a good indicator of the time-dependent variations in cortisol concentrations in plasma following acute stress. However, there is a time lag to reach peak cortisol concentrations in saliva compared to those in plasma, which should be considered when saliva samples are used as the only measure of hypothalamic-pituitary-adrenal axis response to stress in cattle.