Studies addressing factors associated with adverse birth outcomes have almost exclusively been based on hospital statistics. This is a serious limitation in developing countries where the majority of births do not occur within health facilities. This paper examines factors associated with premature deliveries, small baby's size at birth and Caesarean section deliveries in Kenya based on the 1993 Kenya Demographic and Health Survey data. Due to the hierarchical nature of the data, the analysis uses multilevel logistic regression models to take into account the family and community effects. The results show that the odds of unfavourable birth outcomes are significantly higher for first births than for higher order births. Furthermore, antenatal care (measured by frequency of antenatal care visits and tetanus toxoid injection) is observed to have a negative association with the incidence of premature births. For the baby's size at birth, maternal nutritional status is observed to be a predominant factor. Short maternal stature is confirmed as a significant risk factor for Caesarean section deliveries. The observed higher odds of Caesarean section deliveries among women from households of high socioeconomic status are attributed to the expected association between socioeconomic status and the use of appropriate maternal health care services. The odds of unfavourable birth outcomes vary significantly between women. In addition, the odds of Caesarean section deliveries vary between districts, after taking into account the individual-level characteristics of the woman.