Background and objective: Non-adherence is an important issue within severe asthma. Prednisolone and cortisol assays have been proposed as an inexpensive, objective measure of adherence for oral corticosteroid (OCS)-dependent asthmatics, however, little is known about the reliability of these tests. Methods: 41 severe OCS-dependent asthmatics had their prednisolone and cortisol measured during six study visits over a three month time period. Subjects were classed as non-adherent/variably-adherent if they had undetectable prednisolone and/or cortisol >100 nmol/L. Intraclass correlation coefficients (ICCs) were used to assess the test-retest reliability of prednisolone and cortisol, and Gwets AC1 kappa was used to assess the reliability of the adherence classification. Mean change in blood eosinophils for adherent and variably/non-adherent visits were calculated and linear regression with cluster-robust standard errors was used to test for differences. Results: 30 subjects were included in the analysis. Reliability was poor for prednisolone (ICC: 0.43; 95% CI: 0.27, 0.59), and moderate for cortisol (ICC: 0.60; 95% CI: 0.44, 0.74). Using the combined rule, subjects were classified as adherent during 141 (88%) visits, with 21 subjects (70%) adherent during all study visits. The adherence classification had almost perfect reliability (Kappa: 0.84; 95% CI: 0.74, 0.95). Blood eosinophils were decreased by 47 cells/µl (95% CI: 11, 84) during adherent visits but increased by 65 cells/µl (95% CI: -4, 134; Pdifference=0.03) during variably/non-adherent visits. Conclusions: Assessing adherence to maintenance OCS using a simple rule based on prednisolone and cortisol assays is highly reliable and correlated with blood eosinophil changes. Clinicians should have confidence in the results of this rule.