Ethiopia has one of the highest maternal mortality ratios in Africa. Few have examined the quality of labour and delivery (L&D) care in the country. This study evaluated the quality of routine L&D care and identified patient-level and hospital-level factors associated with the quality of care in a subset of government hospitals. This was a facility-based, cross-sectional study using direct non-participant observation carried out in 2016. All mothers who received routine L&D care services at government hospitals (n = 20) in one of the populous regions of Ethiopia, Southern Nations Nationalities and People's Region (SNNPR), were included. Mixed effects multilevel linear regression modeling was employed in two stages using hospital as a random effect, with quality of L&D care as the outcome and selected patient and hospital characteristics as independent variables. Patient characteristics included woman's age, number of previous births, number of skilled attendants involved in care process, and presence of any danger sign in current pregnancy. Hospital characteristics included teaching hospital status, mean number of attended births in the previous year, number of fulltime skilled attendants in the L&D ward, whether the hospital had offered refresher training on L&D care in the previous 12 months, and the extent to which the hospital met the 2014 Ethiopian Ministry of Health standards regarding to resources available for providing quality of L&D care (measured on a 0-100% scale). These standards pertain to availability of human resource by category and training status, availability of essential drugs, supplies and equipment in L&D ward, availability of laboratory services and safe blood, and availability of essential guidelines for key L&D care processes. On average, the hospitals met two-thirds of the standards for L&D care quality, with substantial variation between hospitals (standard deviation 10.9 percentage points). While the highest performing hospital met 91.3% of standards, the lowest performing hospital met only 35.8% of the standards. Hospitals had the highest adherence to standards in the domain of immediate and essential newborn care practices (86.8%), followed by the domain of care during the second and third stages of labour (77.9%). Hospitals scored substantially lower in the domains of active management of third stage of labour (AMTSL) (42.2%), interpersonal communication (47.2%), and initial assessment of the woman in labour (59.6%). We found the quality of L&D care score was significantly higher for women who had a history of any danger sign (β = 5.66; p-value = 0.001) and for women who were cared for at a teaching hospital (β = 12.10; p-value = 0.005). Additionally, hospitals with lower volume and more resources available for L&D care (P-values < 0.01) had higher L&D quality scores. Overall, the quality of L&D care provided to labouring mothers at government hospitals in SNNPR was limited. Lack of adherence to standards in the areas of the critical tasks of initial assessment, AMTSL, interpersonal communication during L&D, and respect for women's preferences are especially concerning. Without greater attention to the quality of L&D care, regardless of how accessible hospital L&D care becomes, maternal and neonatal mortality rates are unlikely to decrease substantially.