The final pattern of neural crest derivatives used to be believed to be the result of unspecified neural crest cells haphazardly entering migratory paths and then receiving cues unique to that path that direct their differentiation. An alternative model, which we have coined the phenotype-directed model, is that neural crest cells are fate-specified first and then select a migratory pathway based on their developmental specification. Support for this model comes from recent studies demonstrating that, at the thoracic level, neural crest cells are specified as melanocyte precursors (melanoblasts) prior to entering the dorsolateral path, and that only melanoblasts have the ability to migrate dorsolaterally. Here we examine two examples of melanocyte patterning in birds that apparently contradict this model. The first is neural crest at the vagal level, where early crest cells migrate dorsolaterally and enter the branchial arches. Despite the fact that these cells migrate dorsolaterally (suggesting that they are melanoblasts), branchial arch-derived neural crest cells fail to differentiate as melanocytes in vitro. These observations suggest that the branchial arch environment may not support the survival or differentiation of melanogenic neural crest cells. The second example is the hyperpigmented Silkie chickens, which exhibit extensive internal pigmentation. The Silkie defect has been linked to a difference in the neural crest migratory environment that potentially causes (or allows) unspecified neural crest cells to undergo melanogenesis in the ventral path. In both of these situations, it appears that the final distribution of pigment cells is controlled by environmental factors, which would contradict the phenotype-directed model. Here we show that the final pattern of melanocytes at the vagal level and in Silkie chickens reflects the migratory behavior of lineage-specified melanoblasts, as predicted by the phenotype-directed model. At the vagal level, the early, dorsolaterally migrating crest cells that colonize the branchial arches are not melanoblasts and are biased against melanogenesis in vitro. Melanoblasts are not specified until later, just prior to a second wave of dorsolateral migration, and although these cells migrate dorsolaterally they do not invade the branchial arches. In Silkie embryos, melanoblasts are specified late and only invade the dorsolateral path after they have been specified. Unlike quail and White leghorn melanoblasts, however, Silkie melanoblasts also migrate ventrally, but again only after they are specified.