OBJECTIVE: To determine whether a non-specialist health worker can accurately undertake audiometry and otoscopy, the essential clinical examinations in a survey of hearing loss, instead of a highly skilled specialist (i.e. ENT or audiologist). METHODS: A clinic-based diagnostic accuracy study was conducted in Malawi. Consecutively sampled participants ≥ 18 years had their hearing tested using a validated tablet-based audiometer (hearTest) by an audiologist (gold standard), an audiology officer, a nurse and a community health worker (CHW). Otoscopy for diagnosis of ear pathologies was conducted by an ENT specialist (gold standard), an ENT clinical officer, a CHW, an ENT nurse and a general nurse. Sensitivity, specificity and kappa (κ) were calculated. 80% sensitivity, 70% specificity and kappa of 0.6 were considered adequate. RESULTS: Six hundred and seventeen participants were included. High sensitivity (>90%) and specificity (>85%) in detecting bilateral hearing loss was obtained by all non-specialists. For otoscopy, sensitivity and specificity were >80% for all non-specialists in diagnosing any pathology except for the ENT nurse. Agreement in diagnoses for the ENT clinical officer was good (κ = 0.7) in both ears. For other assessors, moderate agreement was found (κ = 0.5). CONCLUSION: A non-specialist can be trained to accurately assess hearing using mobile-based audiometry. However, accurate diagnosis of ear conditions requires at least an ENT clinical officer (or equivalent). Conducting surveys of hearing loss with non-specialists could lower costs and increase data collection, particularly in low- and middle-income countries, where ENT specialists are scarce.