Introduction
We examined the degree to which the skills and knowledge of health workers in Ethiopia were retained 18 months after initial maternal and newborn health training and sought to identify factors associated with 18-month skills assessment performance.

Methods
A nonexperimental, descriptive design was employed to assess 18-month skills performance on the topics of Prevent Problems Before Baby Is Born and Prevent Problems After Baby Is Born. Assessment was conducted by project personnel who also received the maternal and newborn health training and additional training to reliably assess health worker performance.

Results
Among the 732 health workers who participated in maternal and newborn health training in 6 rural districts of the Amhara and Oromia regions of Ethiopia (including pretesting before training and a posttraining posttest), 75 health extension workers (78%) and 234 guide team members (37%) participated in 18-month posttest. Among health extension workers in both regions, strong knowledge retention was noted in 10 of 14 care steps for Prevent Problems Before Baby Is Born and in 14 of 16 care steps of Prevent Problems After Baby Is Born. Lower knowledge retention was observed among guide team members in the Amhara region. Across regions, health workers scored lowest on steps that involved nonaction (eg, do not give oxytocin). Educational attainment and age were among the few variables found to significantly predict test performance, although participants varied substantially by other sociodemographic characteristics.