Human embryonic stem (hES) cells are defined by their extensive self-renewal capacity and their potential to differentiate into any cell type of the human body. The challenge in using hES cells for developmental biology and regenerative medicine has been to direct the wide differentiation potential toward the derivation of a specific cell fate. Within the nervous system, hES cells have been shown to differentiate in vitro into neural progenitor cells, neurons, and astrocytes. However, to our knowledge, the selective derivation of any given neuron subtype has not yet been demonstrated. Here, we describe conditions to direct hES cells into neurons of midbrain dopaminergic identity. Neuroectodermal differentiation was triggered on stromal feeder cells followed by regional specification by means of the sequential application of defined patterning molecules that direct in vivo midbrain development. Progression toward a midbrain dopamine (DA) neuron fate was monitored by the sequential expression of key transcription factors, including Pax2, Pax5, and engrailed-1 (En1), measurements of DA release, the presence of tetrodotoxin-sensitive action potentials, and the electron-microscopic visualization of tyrosinehydroxylase-positive synaptic terminals. High-yield DA neuron derivation was confirmed from three independent hES and two monkey embryonic stem cell lines. The availability of unlimited numbers of midbrain DA neurons is a first step toward exploring the potential of hES cells in preclinical models of Parkinson's disease. This experimental system also provides a powerful tool to probe the molecular mechanisms that control the development and function of human midbrain DA neurons.T he isolation of human embryonic stem (hES) cells (1) has stimulated research aimed at the selective generation of specific cell types for regenerative medicine. Although protocols have been developed for the directed differentiation of mouse embryonic stem (ES) cells into therapeutically relevant cell types, such as dopamine (DA) neurons (2, 3), motor neurons (4), and oligodendrocytes (5), the efficient generation of these cell types from hES cells has not yet been reported (6). Earlier studies demonstrating efficient neural differentiation from hES cells (7, 8) have yielded largely ␥-aminobutyric acid (GABA)ergic and glutamatergic neurons with a maximum of 3% DA neurons reported (9). A very recent study (10) reported up to 20% tyrosine hydroxylase (TH)-positive cells from hES cells but did not confirm midbrain DA neuron identity. A bias toward the generation of GABAergic and glutamatergic neurons is also observed in primary rodent and human neural precursor cells isolated from the CNS after expansion in the presence of epidermal growth factor and fibroblast growth factor (FGF). Similar to the work with primary neural precursors, current hES differentiation protocols require expansion of ES-derived neural precursors in FGF2. We have recently shown that extended FGF2 expansion of mouse ES-derived neural precursors selects for forebrain fate...