Objective: To investigate if erythrocyte-methotrexate-polyglutamate (MTX-PG) concentrations in patients with rheumatoid arthritis (RA) are associated with disease activity or adverse events. Methods: We used a longitudinal study design with two cohorts. The derivation cohort included 102 and the validation cohort included 285 patients with RA on MTX. We measured erythrocyte-MTX-PG with 1-5 glutamate residues at 3 months, 6 months and 9 months after MTX start with a liquid chromatography (LC)-mass spectrometry (MS)/MS assay. Outcomes were disease activity score in 28 joints (DAS28) and adverse events. Longitudinal associations of MTX-PG concentrations after 3 months, 6 months and 9 months with DAS28 were tested with a linear mixed model adjusted for age, gender, baseline DAS28, MTX dose and comedication. Results: In the derivation cohort, mean DAS28 decreased from 4.26 (SE=0.14) at baseline to 2.72 (SE=0.13) after 9 months. Thirty per cent of patients in the derivation cohort experienced more than three adverse events after 3 months, which decreased to 18% after 9 months. In the validation cohort, DAS28 and adverse events were comparable with the derivation cohort. In the derivation cohort, MTX-PG1 (β=-0.005), MTX-PG2 (β=-0.022), MTX-PG3 (β=-0.007) and total MTX-PG (β=-0.004) were associated (p<0.05) with lower DAS28 over 9 months. In the validation cohort, MTX-PG2 (β=-0.015), MTX-PG3 (β=-0.010), MTX-PG4 (β=-0.008) and total MTX-PG (β=-0.003) were associated with lower DAS28 over 9 months. None of the MTX-PGs was associated with adverse events. Conclusions: In this first longitudinal study, we showed that an increase in erythrocyte-MTX-PG concentration was associated with a decreased DAS28 over 9 months in two cohorts, and is therefore a potential tool for therapeutic drug monitoring of MTX in RA.