Objective: Cortisol cut-offs can predict requirement for Synacthen stimulation tests (SST). We assessed the performance of a standard cortisol cut-off (375 nmol/L) across the morning and compared this with a time-adjusted cut-off. Design: Retrospective audit. Patients: Community reference set (n=12 550) and SST patients (n=757). Measurements: In the reference population, time-specific cortisol medians were calculated and used to convert cortisol to time-adjusted Multiples of the Median (MoM). In 757 SST patients, the predictive performance of a standard cortisol cut-off (375 nmol/L) and its time-adjusted MoM equivalent were compared. Results: Median cortisol decreased by ~30 nmol/L per hour between 0700 and 1200h. In the reference population, proportions below the 375 nmol/L cut-off increased throughout the morning (range 35%-64%), whereas using the time-adjusted MoM cut-off proportions were consistent (range 46%-50%), with a 17% maximal difference in referral rates between the two cut-offs after 1100h. A similar pattern was noted in the SST cohort. When a cortisol MoM cut-off was used to predict SST success, the excess proportion of patients tested and misclassification rates were lower and more consistent than when the standard cut-off was used. A median cortisol of 375 nmol/L equated to 444 and 313 nmol/L before 0800 and after 1100 h, respectively. Conclusion: The use of a standard cortisol cut-off results in 17% more patients being referred for SST later in the morning. A time-adjusted cortisol cut-off provides consistent and lower referral rates, whilst maintaining similar or better performance than a standard single cut-off in predicting outcome of SST.