This thesis explores the benefits and limitations of traditional evaluations of maternity care, looking specifically at one innovation in service provision, a midwife managed delivery unit. The research undertaken in this thesis can be described in terms of three developmental phases. In phase one, care in a midwife-managed delivery unit is compared with care in a consultant-led labour ward within the framework of a randomised controlled trial. 'Success' is measured in terms of both the clinical aspects of care and as viewed by the women who received this care. Care of women at low obstetric risk in a midwife-managed delivery unit is shown to result in less intervention, greater continuity of carer, more involvement in decision making and greater women's satisfaction with how care was managed. There were no differences in overall satisfaction and the limitations of satisfaction as an outcome measure are discussed. Phases two and three build on the work of the randomised controlled trial. In phase two, perinatal mortality and morbidity data are reviewed through an independent case review of the perinatal deaths and further analysis of the morbidity data. In phase three, the thesis utilises techniques from the discipline of health economics to go beyond the traditional measure of women's views, satisfaction. Willingness to pay and conjoint analysis are used to determine women's preferences, and the strength of these preferences, for different models of maternity service provision.