Background: In order to assess the pharmacokinetic/pharmacodynamic relationship of hydroxychloroquine (HCQ) in patients with systemic lupus erythematosus (SLE), HCQ levels have been measured in whole blood as well as in serum but both methods have never been compared. In addition, cut offs for non-adherence (classically 200ng/ml but also 100 ng/mL) have been established only in whole blood. Objectives: The aims of this study were (1) to compare these two pharmacological approaches, and (2) since it would be very interesting to retrospectively assess severe non-adherence in clinical trials or in large cohort of patients in which only serum samples are usually available, to determine if serum HCQ level cut offs could be established for identification of severe non-adherent patients. Methods: The HCQ and desethylchloroquine (DCQ) levels were measured in serum and whole blood from 573 SLE patients. The risk factors for active SLE (SLEDAI score >4) were identified using multiple logistic regression. HCQ serum level was also measured in 51 non-adherent patients (whole blood HCQ level <200 ng/mL). Results: The mean HCQ and DCQ levels were 916 ± 449 and 116 ± 55 ng/mL in whole blood, respectively; and 469 ± 223 and 63 ± 31 ng/mL in serum, respectively. The mean ratio of serum/whole blood level for HCQ and DCQ were 0.53 ± 0.15 and 0.57 ± 0.21, respectively. A strong positive correlation was found between serum and whole blood levels of HCQ (rho=0.837 [CI95% 0.810-0.860], p<0.0001), and DCQ (rho=0.771 [CI95% 0.736-0.802], p<0.0001). In the multivariate analysis, only corticosteroids (p=0.044), immunosuppressant (p=0.027), HCQ whole blood level (p=0.023) and hemoglobin (p=0.009) were identified as an independent risk factor of active SLE but serum HCQ level was not. Given the mean ratio of serum/whole blood level for HCQ was 0.53, we extrapolated that serum HCQ level cut offs of 106 and 53 ng/mL would correspond to the previously used cut-off of 200 and 100 ng/mL of HCQ in whole blood. Using HCQ serum level cut off of 106 ng/mL, 43 of 51 patients (84%) with blood HCQ levels <200 ng/mL would also have been considered as non-adherent. The positive and negative predictive value of HCQ serum level < 106 ng/ml to detect non-adherence were 96.6% and 63.6%, respectively. Of these 51 patients, 25 patients (49%) exhibited HCQ whole blood concentration below 100 ng/mL. Using HCQ serum level cut off of 53 ng/mL, 23 of 25 patients (92%) with HCQ whole blood level<100 ng/mL, would also have been considered as non-adherent. The positive and negative predictive value of HCQ serum level < 53 ng/ml to detect non-adherence were 82.1% and 90.9%. Conclusion: Our data support the use of whole blood rather than serum as the matrix for drug monitoring of HCQ levels in SLE patients. However, when whole blood is not available, our results support the use of HCQ serum level to assess non-adherence with a cut off of 106 ng/mL corresponding to 200 ng/ml in whole blood