Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) polymerase chain reaction (PCR) on nasopharyngeal swab (NPS), remains the most reliable and practical test to diagnose coronavirus disease 2019 (COVID-19). Current literature is sparse regarding the rates of discordance between NPS and bronchoalveolar lavage (BAL) in patients with cancer. We conducted a retrospective cohort study of adult patients with cancer who had BAL samples tested for SARS-CoV-2 at a comprehensive cancer center. Patients without NPS PCR for SARS-CoV-2 before BAL were excluded. In a cohort of 345 patients, 12% and 17% tested positive for SARS-CoV-2 on NPS and BAL, respectively. There was a 6.3% NPS-/BAL+ discordance rate and a 9.5% NPS+/BAL- discordance rate. Patients with lymphoma (adjusted odds ratio [aOR] = 4.06; P = .007) and Hispanic patients (aOR = 3.76; P = .009) were more likely to have NPS-/BAL+ discordance on multivariate analysis. Among patients with NPS- /BAL- for SARS-CoV-2, an alternate infectious (23%) and a noninfectious etiology (16%) were identified in BAL. Our discordance rates between NPS and BAL were sufficient to recommend BAL in certain patients with cancer with a high clinical suspicion of COVID-19. BAL has value in identifying alternative etiologies of illness in patients with suspected or confirmed COVID-19.