The Multimodal Bio-Medical Systems Workshop (MBM) was held on October 22, 2004 in the Lister Hill Auditorium at the National Library of Medicine, Bethesda, Maryland. It allowed for unique interactions among experts of life sciences and physical sciences working on various aspects of multimodal biomedical systems research, whose goals are to collect, process, analyze, integrate, model, synthesize, visualize, and simulate vast amounts of heterogeneous multi-modal, multi-scale data for emerging real world applications in life science. The workshop targeted both the micro/nano level and the macro level of biomedical systems. At the micro/nano level, unlike alphanumeric genetics and genomics data, multimodality biological data are increasingly generated by high throughput or high content data acquisition devices, including a broad spectrum of microscopes (e.g., fluorescence, light, optical, confocal, two-photon laser scanning, time-lapse, laser capture dissection, and electronic), biosensors, and microarrays. The pragmatic goal is to translate scientific discovery into clinical practice, including diagnosis, prognosis, therapeutic prevention, repair, drug development, and disease management, in an effective manner. During the workshop, the attendees reached considerable consensus to address the following items as the selected grand challenges of multimodal biomedical systems: to allow early detection of where and when an infectious disease outbreak occurs, whether it is naturally occurring or man-made, in real time; to develop multidimensional drug profiling data bases to facilitate drug discovery and to identify biomarkers for diagnosis and monitoring the progress of individual disease treatments; to connect activities and events derived from cellular processes to high level cognition; and to support personalized medical care and clinical decision for patients.