The clinical predictors of positive sputum culture have not been previously reported in hospital-acquired pneumonia (HAP) and data on yield of sputum culture in this setting are scant. Current IDSA guidelines for HAP recommend non-invasive sputum sampling, though the data for this practice are limited. We assessed the yield of sputum culture in HAP cases at an academic medical center from 1/2007–7/2013. HAP cases were identified by ICD-9 codes for bacterial pneumonia and all cases were validated by chart review. Our cohort had 1,172 hospitalizations with a HAP diagnosis. At least 1 sputum specimen was collected non-invasively and sent for bacterial culture after hospital day 2 and within 7 days of HAP diagnosis in 344 of these hospitalizations (29.4%), with a total of 478 sputum specimens, yielding 63 (13.2%) positive, 109 (22.8%) negative, and 306 (64.0%) contaminated cultures (>10 epithelial cells/hpf). Significant predictors of a positive sputum culture were chronic lung disease (RR 2.0, 95% confidence interval 1.2–3.4) and steroid use (RR 1.8, 95% confidence interval 1.1–3.2). The most commonly identified organisms were Gram-negative rods not further speciated (25.9%), Staphylococcus aureus (21.0%) and Pseudomonas aeruginosa (14.8%). The ease of obtaining a sputum sample combined with the prevalence of commonly drug-resistant organisms suggests that sputum culture in HAP is a potentially useful non-invasive diagnostic technique.