BACKGROUND: In low-income countries, surgical site infections (SSIs) are a very frequent form of hospital-acquired infection. Surveillance is an important method for controlling SSI but it is unclear how this can best be performed in low-income settings. AIM: To examine the epidemiological characteristics of various components of an SSI surveillance programme in a single Kenyan hospital. METHODS: The study assessed the inter-observer consistency of the surgical wound class (SWC) and American Society of Anesthesiologists (ASA) scores using the kappa statistic. Post-discharge telephone calls were evaluated against an outpatient clinician review 'gold standard'. The predictive value of components of the Centers for Disease Control and Prevention - National Healthcare Safety Network (CDC-NHNS) risk index was examined in patients having major obstetric or gynaecological surgery (O&G) between August 2010 and February 2011. FINDINGS: After appropriate training, surgeons and anaesthetists were found to be consistent in their use of the SWC and ASA scores respectively. Telephone calls were found to have a sensitivity of 70% [95% confidence interval (CI): 47-87] and a specificity of 100% (95% CI: 95-100) for detection of post-discharge SSI in this setting. In 954 patients undergoing major O&G operations, the SWC score was the only parameter in the CDC-NHNS risk index model associated with the risk of SSI (odds ratio: 4.00; 95% CI: 1.21-13.2; P = 0.02). CONCLUSIONS: Surveillance for SSI can be conducted in a low-income hospital setting, although dedicated staff, intensive training and local modifications to surveillance methods are necessary. Surveillance for post-discharge SSI using telephone calls is imperfect but provides a practical alternative to clinic-based diagnosis. The SWC score was the only predictor of SSI risk in O&G surgery in this context.