BACKGROUND: Our study aimed to assess the effect of Helping Mothers Survive Bleeding after Birth on knowledge and skills of health workers and whether such effect varies by health workers characteristics. METHODS: Nested in a cluster-randomised trial to assess the effect of the training on health outcomes, we assessed changes in knowledge and simulated skills in 61 facilities. The assessments were done i) before, ii) immediately-after training session and iii) at 10-month follow-up for subset of health-workers of implementation facilities as defined by the trial. We used a self-administered questionnaire and Objective Structures Clinical Examinations to assess three skill sets: Active Management of Third Stage of Labour, removal of retained placenta and management of severe postpartum haemorrhage. We computed summary statistics and used the paired t-test to assess change of knowledge and skills immediately post-training and at 10-month follow-up. Linear regression was done to assess association of scores and health worker characteristics. RESULTS: Of the 636 health workers included, 606 (96.7%) and 591 (91.4%) completed the knowledge and skills assessments, respectively. Majority of the participants (68%) were nurse-midwives. Knowledge scores increased by 15 percentage-points from 77.5% to 93% (95% CI 14.3, 16.3, p-value <0.000), and skills scores by 47 percentage-points (95% CI 46.5, 49.2, p-value <0.000) from 37.5% to 83%. There was a 4.0% decline of skills at 10-month follow-up. The decline was higher in auxiliary staff (-11.8%) and least in nurse-midwives (-2.1%) p-value <0.001. Health workers who assisted less than 5 deliveries in the last month, those who never attended postpartum haemorrhage in-service training and profession experience >8 years were associated with lower mean skill change immediately post-training. CONCLUSION: Our study supports the potential of the Helping Mothers Survive Bleeding after Birth training to increase knowledge and skills of postpartum haemorrhage among all professional groups. Auxiliary staff benefited most from the training but also showed higher skill decline at 10-month. Our study highlights the importance to disaggregate knowledge and skills by health workers characteristics.