Background: In resource limited-settings, timely tuberculosis (TB) diagnosis depends upon referral of sputum samples from non-diagnostic to diagnostic facilities for examination. The TB programme data for 2018 suggested losses in Mpongwe District’s sputum referral cascade.Aim: This study aimed to identify the referral cascade stage where loss of sputum specimen occurred.Setting: Primary health care facilities in Mpongwe District, Copperbelt Province, Zambia.Methods: Data were retrospectively collected from one central laboratory and six referring health facilities between January and June 2019, using a paper-based tracking sheet. Descriptive statistics were generated in SPSS version 22.Results: Of the 328 presumptive pulmonary TB patients found in presumptive TB registers at referring facilities, 311 (94.8%) submitted sputum samples and were referred to the diagnostic facilities. Of these, 290 (93.2%) were received at the laboratory, and 275 (94.8%) were examined. The remaining 15 (5.2%) were rejected for reasons such as ‘insufficient sample’. Results for all examined samples were sent back and received at referring facilities. Referral cascade completion rate was 88.4%. Median turnaround time was six days (IQR = 1.8).Conclusion: Losses in the sputum referral cascade for Mpongwe District mainly occurred between dispatch of sputum samples and receipt at diagnostic facility. Mpongwe District Health Office needs to establish a system to monitor and evaluate the movement of sputum samples along the referral cascade to minimize losses and ensure timely TB diagnosis.Contribution: This study has highlighted, at primary health care level for resource limited settings, the stage in the sputum sample referral cascade where losses mainly occur.