BACKGROUND: Each year an estimated four million neonates die, the majority in the first week of life. One of the major causes of death is sepsis. Proving the incidence and aetiology of neonatal sepsis is difficult, particularly in resource poor settings where the majority of the deaths occur. METHODS: We conducted a three year observational study of clinically diagnosed early onset (<7 days of age) neonatal sepsis (EONS) in infants born to mothers following antenatal care at the Shoklo Malaria Research Unit clinic in Maela camp for displaced persons on the Thailand-Myanmar border. Episodes of EONS were identified using a clinical case definition. Conventional and molecular microbiological techniques were employed in order to determine underlying aetiology. RESULTS: From April 2009 until April 2012, 187 infants had clinical signs of EONS, giving an incidence rate of 44.8 per 1000 live births (95% CI 38.7-51.5). One blood culture was positive for Escherichia coli, E. coli was detected in the cerebrospinal fluid specimen in this infant, and in an additional two infants, by PCR. Therefore, the incidence of bacteriologically proven EONS was 0.7 per 1000 live births (95% CI 0.1-2.1). No infants enrolled in study died as a direct result of EONS. CONCLUSION: A low incidence of bacteriologically proven EONS was seen in this study, despite a high incidence of clinically diagnosed EONS. The use of molecular diagnostics and nonspecific markers of infection need to be studied in resource poor settings to improve the diagnosis of EONS and rationalise antibiotic use.