Neural crest-derived melanophores form species-specific patterns in the dermis of amphibian embryos. Melanophore patterns may be generated by one of two general mechanisms: 1) pigment cell precursors disperse throughout the embryo, with melanophores differentiating in certain regions due to environmental cues, or 2) melanoblasts may localize in different regions as a result of a hierarchy of tissue affinities. Both of these mechanisms have been proposed to be responsible for the dorso-ventral patterning of melanophores in Xenopus laevis. We have reexamined the distribution of melanoblasts in X. laevis and Taricha torosa using the dopa (3,4-dihydroxyphenylalaninel-staining technique. We have found that many of the dopa-positive cells identified as melanoblasts by some researchers are actually not derived from the neural crest: 1) dopa-positive cells in I: torosa were identified in the transmission electron microscope to be either leukocytes or erythrocytes, 2) in X. laevis dopa-positive cells are found between the ectoderm and somites where neural crest cells are not found, and 3) X. Eaevis embryos surgically depleted of neural crest have dopa-staining patterns identical to control embryos. Melanoblasts are apparently not found in the ventralmost regions of early 2: torosa and X. laevis embryos, providing additional evidence for the role of differential tissue affinities in directing the formation of embryonic pigment cell patterns.