Rheumatoid arthritis (RA) is characterised by clinical joint swelling and elevation of acute phase reactant levels, typically measured by the C-reactive protein (CRP). Clinical and inflammatory responses are usually concordant, except for inhibition of IL-6, which often disproportionally reduces the CRP due to direct inhibition of its hepatic production. We investigated whether pre-treatment CRP is a useful marker that can guide a preferential treatment choice towards IL-6 inhibition.Data of 1126 treatment courses with tocilizumab (TCZ; early RA), 250 courses of rituximab (RTX; established RA) and 249 courses of methotrexate (MTX; established RA) were analysed. We compared clinical disease activity index (CDAI) values and change along 24 weeks' follow-up to CRP values at baseline or its early change. We validated the results using data from a separate TCZ trial in early RA.CRP levels in the TCZ group on average dropped by 74% within 4 weeks. Patients who attained CDAI remission at 24 weeks on TCZ had the highest baseline CRP levels while patients in high disease activity had the lowest; this association was reverse in the RTX and MTX groups. TCZ patients who achieved remission at 24 weeks showed the largest reductions of CRP levels by week 4 compared with those reaching higher disease activity states. Early CRP non-response was indicative of a risk of not achieving clinical treatment goals (p=0.038).Baseline CRP appears to have a positive association with reaching the therapeutic target on TCZ treatment, but is a negative predictor for RTX and MTX. Patients on TCZ without an early CRP response have a lower chance of achieving remission. CRP and its early course may inform, to some extent, the estimation of potential therapeutic success in patients with RA.