Early warning scores are points-based or colour-coded systems used to detect changes in physiological parameters and prompt earlier recognition and management of deteriorating patients. Vital signs recorded within a coloured zone corresponding to degree of derangement ('trigger') should prompt an action. The report of the UK Confidential Enquiry into Maternal and Child Health recommends the use of modified versions in the obstetric population. Currently, there is limited research into the effects of early warning scores in low-resource settings where maternal mortality remains high, and there is a need for low-cost, simple methods to reduce this. A modified obstetric early warning system (MOEWS) was introduced for parturients who had undergone surgical intervention at Felege Hiwot Referral Hospital, a tertiary centre in Bahir Dar, Ethiopia. A guideline was developed to accompany the MOEWS, together with training of healthcare workers. Prior to introduction, the quality of postoperative monitoring was assessed through retrospective case note review. This was reassessed at 8 months and 11 months postimplementation, with assessment of response to 'triggers'. A questionnaire and qualitative interviews were undertaken to establish views of healthcare workers on its acceptability and usability. Recording of postoperative vital signs improved with the implementation of the MOEWS and was sustained at both monitoring periods. The number of patients with vital signs within the coloured zones ('trigger') was reduced, although documented action to these remained low. Staff were positive towards the MOEWS, its impact on patient care and felt confident using the system. The introduction of a MOEWS in an Ethiopian referral hospital in this study appeared to improve the monitoring of postoperative patients. With modifications to suit the setting and senior clinician involvement, coupled with regular training, the early warning score is a feasible and acceptable tool to cope with the unique demands faced in this low-resource setting.