Aims: Tocilizumab has a direct effect on inflammatory markers. Therefore, composite measures for disease activity assessment in rheumatoid arthritis (RA) using these inflammatory markers may not be suitable for tocilizumab treatment. We used a modelling approach to describe the tocilizumab exposure-response relationship and to investigate the different dynamics of the individual components of the routinely used continuous composite measures. Methods: Pharmacokinetic (PK), clinical and laboratory data were obtained from a prospective, observational, single-centre study involving 35 subjects with RA treated with intravenous tocilizumab. A population PK/pharmacodynamic analysis was performed using nonlinear mixed effects models. Results: The population for model development comprised 1086, 1083 and 1082 observations calculated with the disease activity score based on 28 joint (DAS28) and the simplified and clinical disease activity scores (SDAI, CDAI). The tocilizumab exposure-response relationship was described with an indirect-response model. Two main groups of individual components were identified based on their different dynamics under tocilizumab treatment: (i) tender and swollen joint counts and patient and evaluator global assessment showed a slower decrease of their baseline value (half-life: 4.6 weeks, RSE: 24%) and the need for higher serum drug concentration (EC50: 4.60 μg/mL, RSE: 103%, IIV: 359%) than (ii) C-reactive protein and erythrocyte sedimentation rate (half-life: 2.3 weeks, RSE 19%; EC50: 0.878 μg/mL, RSE: 41%, IIV: 238%). Conclusion: Our study confirms a different dynamics of the individual components of the most frequently used continuous composite measures under tocilizumab treatment which should be taken into account to avoid misassessment of disease activity.