Patients with undifferentiated arthritis (UA) have clinically apparent inflammatory arthritis but no evident diagnosis or classification. Considering UA as a definition 'per exclusionem' implies that the population designated by this term is affected by changes in the way other diseases, eg, rheumatoid arthritis (RA), are classified or diagnosed. Current treatment recommendations for UA are largely similar to recommendations for RA. The recommendations are based on the idea that UA is an early stage of RA and on literature generated in the 2000s before the development of the 2010 classification criteria for RA. However, conventional UA (so-called '1987-UA') is presumably different than contemporary UA ('2010-UA'). Strikingly, there are no randomised placebo-controlled trials done on '2010-UA,' and this poses questions on whether the recommendations for UA are still valid. In this absence, we assume that treatment recommendations from '1987-UA' can be extrapolated to '2010-UA' if (1) essential patient characteristics are the same, (2) long-term outcomes are similar, (3) prognostic factors are largely the same, and (4) there are indications from research other than placebo-controlled randomized clinical trials (RCTs) that disease modifying antirheumatic drug (DMARD) treatment in 2010-UA is effective. We evaluate these requirements one by one based on the literature on 2010-UA. This reveals that 2010-UA is milder in initial presentation and disease outcomes than 1987-UA. Today's UA population is >
95% anticitrullinated protein antibody-negative, presents with mono- or oligoarthritis, frequently achieves spontaneous remission, and rarely progresses to RA. We suggest that 2010-UA is a distinct patient group within the early arthritis spectrum, requiring additional research, after which recommendations may need to be updated.