Abstract Background We assessed coverage of symptom screening and sputum testing for tuberculosis (TB) in hospital outpatient clinics in Ghana. Methods In a cross-sectional study, we enrolled adults (≥18 years) exiting the clinics reporting ≥1 TB symptom (cough, fever, night sweats or weight loss). Participants reporting a cough ≥2 weeks or a cough of any duration plus ≥2 other TB symptoms (per national criteria) and those self-reporting HIV-positive status were asked to give sputum for testing with Xpert MTB/RIF. Results We enrolled 581 participants (median age 33 years [IQR: 24–48], 510/581 [87.8%] female). The most common symptoms were fever (348, 59.9%), chest pain (282, 48.5%) and cough (270, 46.5%). 386/581 participants (66.4%) reported symptoms to a healthcare worker, of which 157/386 (40.7%) were eligible for a sputum test per national criteria. Only 31/157 (19.7%) had a sputum test requested. Thirty-two additional participants gave sputum among 41 eligible based on positive HIV status. In multivariable analysis, symptom duration ≥2 weeks (adjusted odds ratio [aOR] 6.99, 95% confidence interval [CI] 2.08–23.51) and previous TB treatment (aOR: 6.25, 95% CI: 2.24–17.48) were the strongest predictors of having a sputum test requested. 6/189 (3.2%) sputum samples had a positive Xpert MTB/RIF result. Conclusion Opportunities for early identification of people with TB are being missed in health facilities in Ghana.