Non-adherence to disease-modifying antirheumatic drugs (DMARDs) hampers the targets of rheumatoid arthritis (RA) treatment, obtaining low disease activity and decreasing radiological progression. This study investigates if, and to what extent, non-adherence to treatment would lead to a higher 28-joint count disease activity score (DAS28) in the first year after diagnosis. Adult patients from an ongoing cohort study on treatment adherence were selected if they fulfilled the EULAR/ACR2010 criteria for RA, and were to start with their first DMARDs. Clinical variables were assessed at baseline and every 3 months. Non-adherence was continuously electronically measured and was defined as the proportion of days with a negative difference between expected and observed openings of the medication container out of the 3-month period before DAS28 measurement. Generalized linear mixed models were used to investigate whether the DAS28 related to non-adherence. Covariates included were age, sex, baseline DAS28, rheumatoid factor positivity, anti-cyclic citrullinated peptide antibodies (ACPA) positivity, anxiety, depression, weeks of treatment, number of DMARDs used, education level, use of subcutaneous methotrexate and biological use. One hundred and twenty patients met the inclusion criteria for RA. During the study period 17 patients became lost to follow-up. There was a decline in adherence over time for all DMARDs except for prednisone. Non-adherence is a predictor of disease activity in the first 6 months of therapy, adjusted for weeks of treatment, baseline DAS28, and baseline anxiety. Non-adherence relates to disease activity. Therefore, interventions towards enhancing adherence can improve disease outcome.