We aimed to assess whether patient-physician discordance regarding disease activity affects T2T-implementation and clinical outcomes in rheumatoid arthritis (RA). This was an analysis of the 2-year T2T-guided trial Care in early RA (CareRA). During year 1, DMARD escalations were mandated by the protocol when DAS28-CRP was >3.2. During year 2, treatment was at the rheumatologists' discretion. At each visit we assessed T2T-implementation, defined as escalating DMARDs if DAS28-CRP >3.2. Patient-physician discordance was defined by the discordance score (DS), a weighted difference between patient-reported and clinical/laboratory outcomes. Using generalised linear mixed models and multilevel mediation analysis, we studied the association between time-varying DS, T2T-implementation and the odds of remission (SDAI ≤3.3), physical functioning (HAQ-score), and radiographic progression at year 2. Over 2 years, 379 patients were assessed at 3129 follow-up visits. On 445 (14%) of these visits, DAS28-CRP was >3.2, and DMARDs were escalated in 217/445 (49%) of such cases. T2T-implementation declined over time and was consistently lower during the second year (year 1: 57-66%; year 2: 17-52%). Higher DS over time was negatively associated with remission and physical functioning at year 2, partly mediated by a lower proportion of T2T-adherent visits. No such association was found for radiographic progression. Even in a trial setting, T2T was applied on only around 50% of visits. T2T was less likely to be implemented with increasing patient-physician discordance regarding disease activity, which was in turn associated with less remission and worse functional outcome, but not with radiographic progression.