Inability to reduce neonatal and maternal mortality in poor countries is sometimes blamed on a lack of contextual knowledge about care practices and care-seeking behavior. There is a lack of knowledge about how to translate formative research into effective interventions to improve maternal and newborn health. We describe the findings of formative research and how they were used to inform the development of such an intervention in rural Nepal. Formative research was carried out in four parts. Part 1 involved familiarization with the study area and literature review, and parts 2, 3 and 4 involved community mapping, audit of health services, and qualitative and quantitative studies of perinatal care behaviors. Participatory approaches have been successful at reducing neonatal mortality and may be suitable in our context. Community mapping and profiling helped to describe the community context, and we found that community-based organizations often sought to involve the Female Community Health Volunteer in community mobilization. She was not routinely conducting monthly meetings and found them difficult to sustain without support and supervision. In health facilities, most primary care staff were in post, but doctors and staff nurses were absent from referral centers. Mortality estimates reflected under-reporting of deaths and hygiene and infection control strategies had low coverage. The majority of women give birth at home with their mother-in-law, friends and neighbors. Care during perinatal illness was usually sought from traditional healers. Cultural issues of shyness, fear and normalcy restricted women's behavior during pregnancy, birth and the postpartum period, and decisions about her health were usually made after communications with the family and community. The formative research indicated the type of intervention that could be successful. It should be community-based and should not be exclusively for pregnant women. It should address negotiations within families, and should tailor information to the needs of local groups and particular stakeholders such as mothers-in-law and traditional healers. The intervention should not only accept cultural constructs but also be a forum in which to discuss ideas of pollution, shame and seclusion. We used these guidelines to develop a participatory, community-based women's group intervention, facilitated through a community action cycle. The success of our intervention may be because of its acceptability at the community level and its sensitivity to the needs and beliefs of families and communities.