We aimed to retrospectively evaluate retention rate and causes of discontinuation of JAKi in rheumatoid arthritis (RA) patients with particular regards to difficult-to-treat subgroups. The diffusion of Janus kinase inhibitors (JAKi) for the treatment of RA has rapidly increased in recent years due to their effectiveness, even in difficult-to-treat subgroups of patients. After the publication of the Oral Surveillance study, the labelling of JAKi was modified, advising against their use in elderly patients and those at risk for cardiovascular events and malignancies. Demographic, clinical, serological and therapeutic characteristics of RA patients treated with JAKi were recorded, including smoking habit and comorbidities. Three hundred and thirty consecutive RA patients were enrolled in the study. Among them, 50.3% patients had previously failed at least two biologic DMARDs. Risk factors for the use of JAKi were reported in 75.5% of patients, 41.5% of them were older than 65 years, 37.6% had smoked, while 48.8% had increased cardiovascular or cancer risk. Anticitrullinated peptide antibodies (ACPA) and combination therapy with conventional synthetic DMARDs were associated with a longer drug persistence and ACPA remained independently associated to a higher retention rate of JAKi also in the subgroup of difficult-to-treat patients. In conclusion, our study supports the clinical effectiveness of JAKi in RA, even in the multi-failure subgroup of patients, where the risk/benefit ratio overcomes the safety risk. The presence of ACPA and the concurrent use of + cs-DMARD may increase the survival on JAKi in the long term.