The initial, site-specific colonization of secondary organs by blood-borne cancer cells appears to be mediated by endothelial cell adhesion molecules. These molecules are part of the organ-specific microvascular phenotype and are regulated through complex interactions of the endothelium with the extracellular matrix (e.g., distinct matrix macromolecules and growth factors). They are induced in vitro by growing 'unspecific' (large vessel) endothelial cells on extracts of organ-specific biomatrices. In many respects, these molecules are similar to the various classes of chemically different adhesion molecules that regulate lymphocyte traffic, but are believed to be distinct from the inducible adhesion molecules that govern leukocyte adhesion during acute episodes of inflammation. Biochemical and biophysical data indicate that preference of tumor cell adhesion to organ-specific microvascular endothelium may not require qualitative differences of such homing receptors between endothelia, but may be explained on the basis of quantitative receptor differences as well as differences of receptor avidity. Following adhesion, the metastatic cascade proceeds by the establishment of metabolic conduits between the endothelium and adherent tumor cells. This heterotypic coupling represents an early step in the extravasation of cancer cells from the microvasculature, initiating endothelial cell retraction from its basement membrane and recanalization around the arrested tumor cell. These events, together with local growth promoting effects exerted by the metastasized organ, are believed to provide the basis for Paget's 'seed and soil' hypothesis of metastasis.