Our aim was to determine whether clinician-identified poor compliance is useful in identifying, from among adults with severe asthma, patients with characteristics likely to put them at risk of adverse outcomes. Patients with severe asthma (previous hospital admissions and/or prescribed step 4-5 treatment according to British Thoracic Society guidelines) considered by clinicians to be either compliant (C, N = 41) or poorly compliant (PC, N = 92) with aspects of their recommended management (attendance at appointments, taking medication, and monitoring asthma) provided data on symptoms, health service use, medication, self-management practices, physical and psychological comorbidities, and sociodemographic/socioeconomic characteristics. Cross-sectional univariate analyses were used to examine whether the groups differed with respect to self-reported indicators of asthma morbidity and self-management. Logistic regressions were additionally used to explore psychosocial factors independently associated with patients being identified as PC. Compared with C patients, PC patients had significantly poorer self-reported asthma control in terms of medication use, symptoms, time off work, asthma-specific quality of life, primary care visits, emergency attendances, and hospital admissions. This was coupled with poorer self-management practices. Patients identified as PC also had higher levels of physical and psychological comorbidities, were younger, and faced more difficult social and economic circumstances. We identified significant psychological (anxiety) and social (younger age, not working, number of benefits, adverse family circumstances) factors independently associated with patients being identified as PC. Among adults with severe asthma, clinician-assessed poor compliance was useful in distinguishing between two groups that differed significantly in terms of asthma morbidity indicators, self-management practices, and psychosocial characteristics, which have been previously shown to be associated with hospital admissions, near-fatal attacks, and fatal asthma. We conclude that clinician-assessed poor compliance is a useful marker for identifying patients at risk of these adverse outcomes.