The central nervous system (CNS) of Xenopus is derived from three of four tiers of blastomeres of the 32-cell embryo, and each blastomere in these tiers produces a characteristic number of primary spinal neurons. The C-tier blastomeres constitute the boundary between those that contribute to the CNS (A-, B-, and C-tiers) and those that do not (D-tier). To test whether the neural lineages descended from the C-tier are established by intrinsic information or by cell-cell interactions, single B-tier blastomeres were deleted and the lineage of their C-tier neighbors mapped. The contributions of C-tier blastomeres to subdivisions of the CNS and to specific spinal neurons were significantly reduced. Contributions of these blastomeres to other tissues were mostly normal, indicating that those C-tier progeny that no longer contribute to CNS are distributed in small numbers throughout the rest of the clone. To test whether the changes in neural lineages after B-tier deletions were the result of the C-tier blastomere changing position, contacting new neighbors, or losing contact with inductive B-tier neighbors, intact embryos were transiently dissociated within their vitelline membranes at different time points prior to the midblastula transition. This treatment disrupted cell-cell contact, but not gap junction-mediated dye coupling or the positions of neighboring cells. C-tier CNS lineages were reduced as after deletion of the B-tier neighbor, suggesting that the neural fate of C-tier cells depends upon specific B-tier interactions. To determine whether these interactions occurred specifically between B-tier and C-tier neighbors, barriers were inserted transiently between individual B/C pairs; similar reductions in C-tier CNS lineages were observed. These data demonstrate that an animal-to-vegetal, contact-dependent signal passes from B-tier to C-tier blastomeres and is required for the normal C-tier contribution to the CNS. This cell-cell interaction occurs many hours before the onset of zygotic transcription or neural induction and may bias the field of cells that can respond to neural induction.