To advance knowledge about the treatment of addiction among pregnant women and other women of childbearing age, investigators must adhere to the requirements of a strict experimental research design while concurrently providing clinical services. This means that researchers must address a variety of difficult questions, including the following: Was the sample large enough? Were the criteria for subject inclusion and exclusion well defined? Did the process of recruitment result in a sample that could be generalized to a larger population, or was the sample biased in some way? Was assignment to groups clearly random? What was the attrition rate? Was attrition the same in both experimental and comparison groups? Did baseline measures collect enough information to permit a description of the facts that were associated with attrition in each group? Was the attrition rate so high that the retained sample had special characteristics? If so, what were these features? This chapter highlights several problems related to these questions, describes the difficulties that investigators have faced in meeting clinical and research challenges to date, and suggests strategies for overcoming some obstacles. In establishing the Perinatal-20 project, the National Institute on Drug Abuse took an informed first step in organizing a substantial research effort to investigate treatment modalities that incorporate services specific to the needs of substance-abusing women who have children. This initial effort has resulted in a beginning knowledge base that can be used to refine and expand future treatment efforts. Even the issue of the "study unit" for this population is evolving. Today's researchers are attempting to determine whether the mother alone or the mother along with her dependent children constitutes the study unit. This question also has led professionals in the field to examine a range of specific outcome priorities, and investigators just now are beginning to determine exactly what needs to be evaluated in gauging the effectiveness of treatment. Is success measured on the basis of the woman's progress with abstinence alone, or does it also include her role with her children? Is it determined on the basis of her relationship with her children or the children's growth and development? Compared with providing services for and studying single adult subjects, developing treatment for women and their children presents researchers with a more complex task and requires expanded clinical services (Gallagher 1990, pp. 540-559). As in most fields of study, initial research data in substance abuse treatment for pregnant and parenting women are derived from samples of convenience, as described above. To put this information in perspective, future research will require a wider and more representative spectrum of the population. Furthermore, tensions between clinical needs and research requirements must be considered in advance, and methods for relaxing these tensions will be critical to the success of future efforts. For example, members of both the research and clinical staff teams must be absolutely clear about the study design and the requirements of reliable research. Where possible, potential ambiguities about group assignment, project services, subjects' responsibilities, and so forth must be incorporated into subject consent forms so that the subjects also are apprised of potential problems and their solutions. A final caution to future investigators is to be aware of the economic, physical, and personnel limitations of the range of treatment services that can be provided in a research demonstration study involving this population. Because of these limitations and the extensive range of services the subjects of the studies require, treatment components must be discrete and carefully defined to prevent programs from becoming impractically diverse and unclear. Research goals must be attainable and measurable.(ABSTRACT TRUNCATED)