Objective To compare incidences of neuroinflammatory events, including demyelinating disease (DML), inflammatory polyneuropathies (IPN) and multiple sclerosis (MS), in patients with rheumatoid arthritis (RA) or spondyloarthritis (SpA; including psoriatic arthritis) starting a tumour necrosis factor inhibitor (TNFi), investigating whether monoclonal TNFi antibodies (other TNFis (oTNFis)) confer higher risk than etanercept. Methods This is an observational cohort study including patients from the five Nordic countries starting a TNFi in 2001-2020. Time to first neuroinflammatory event was identified through register linkages. We calculated crude incidence rates (cIR) per 1000 person-years and used multivariable-adjusted Cox regression to compare incidences of neuroinflammatory events overall and for DML, IPN and MS with oTNFi versus etanercept. We further examined individual TNFis and indications. Results 33 883 patients with RA and 28 772 patients with SpA were included, initiating 52 704 and 46 572 treatment courses, respectively. In RA, we observed 135 neuroinflammatory events (65% DML) with cIR of 0.38 with oTNFi and 0.34 with etanercept. The HR of oTNFi versus etanercept was 1.07 (95% CI 0.74 to 1.54) for any neuroinflammatory event, 0.79 (95% CI 0.51 to 1.22) for DML, 2.20 (95% CI 1.05 to 4.63) for IPN and 0.73 (95% CI 0.34 to 1.56) for MS. In SpA, we observed 179 events (78% DML) with cIR of 0.68 with oTNFi and 0.65 with etanercept. The HR for any neuroinflammatory event, DML, IPN and MS was 1.06 (95% CI 0.75 to 1.50), 1.01 (95% CI 0.68 to 1.50), 1.28 (95% CI 0.61 to 2.69) and 0.94 (95% CI0.53 to 1.69), respectively. Conclusion The cIRs of neuroinflammatory events are higher in SpA than in RA, but the choice of specific TNFi does not seem to play an important role in the risk of neuroinflammatory events.