Embryonic stem (ES) cells have tremendous potential as a cell source for cell-based therapies. Realization of that potential will depend on our ability to understand and manipulate the factors that in¯uence cell fate decisions and to develop scalable methods of cell production. We compared four standard ES cell differentiation culture systems by measuring aspects of embryoid body (EB) formation ef®ciency and cell proliferation, and by tracking development of a speci®c differentiated tissue typeÐbloodÐusing functional (colony-forming cell) and phenotypic (Flk-1 and CD34 expression) assays. We report that individual murine 1 1 1 1 1 1 1 1 1 1 ES cells form EBs with an ef®ciency of 2 2 2 2 2 2 2 2 2 2 42 9%, but this value is rarely obtained because of EB aggregationÐa process whereby two or more individual ES cells or EBs fuse to form a single, larger cell aggregate. Regardless of whether EBs were generated from a single ES cell in methylcellulose or liquid suspension culture, or aggregates of ES cells in hanging drop culture, they grew to a similar maximum cell number of 28,000 9,000 cells per EB. Among the three methods for EB generation in suspension culture there were no differences in the kinetics or frequency of hematopoietic development. Thus, initiating EBs with a single ES cell and preventing EB aggregation should allow for maximum yield of differentiated cells in the EB system. EB differentiation cultures were also compared to attached differentiation culture using the same outputs. Attached colonies were not similarly limited in cell number; however, hematopoietic development in attached culture was impaired. The percentage of early Flk-1 and CD34 expressing cells was dramatically lower than in EBs cultured in suspension, whereas hematopoietic colony formation was almost completely inhibited. These results provide a foundation for development of ef®cient, scalable bioprocesses for ES cell differentiation, and inform novel methods for the production of hematopoietic tissues.