The activity proposed herein is to make a database of research questions in cancer (RQDB). The idea is that each record will represent exactly one research question. For example, """"""""Is there a necessary causative agent for AIDS?"""""""" For this particular question, the research community has a current consensus answer (yes), but for many open RQs either answers are unknown or conflicting: Does emotional state modify cancer development? Are pesticides, insecticides and other chemicals applied to food (like ALAR) significant factors inhuman carcinogenesis? Do singlet oxygen and free radicals play significant roles in cancer induction? RQDB will contain: active questions, (each with a unique number), current candidate answers (with pointers to appropriate records in CANCERLIT and MEDLINE), indexing (to facilitate retrieval), links to broader and narrower questions, and open/closed assessments for each question. In Phase I, the work will draw on the approximately 173 Oncology Overview publications of the NCI, each of which was composed by a distinguished reviewer, but none of which are available in a computer database. It is hoped that RQDB will give a user a more balanced, complete and direct overview of what is known than is now possible via direct access to MEDLINE and CANCERLIT.