The regulation of development by Hox proteins is important in the evolution of animal morphology, but how the regulatory sequences of Hox-regulated target genes function and evolve is unclear. To understand the regulatory organization and evolution of a Hox target gene, we have identified a wing-specific cis-regulatory element controlling the knot gene, which is expressed in the developing Drosophila wing but not the haltere. This regulatory element contains a single binding site that is crucial for activation by the transcription factor Cubitus interruptus (Ci), and a cluster of binding sites for repression by the Hox protein Ultrabithorax (UBX). The negative and positive control regions are physically separable, demonstrating that UBX does not repress by competing for occupancy of Ci-binding sites. Although knot expression is conserved among Drosophila species, this cluster of UBX binding sites is not. We isolated the knot wing cis-regulatory element from D. pseudoobscura, which contains a cluster of UBX-binding sites that is not homologous to the functionally defined D. melanogaster cluster. It is, however, homologous to a second D. melanogaster region containing a cluster of UBX sites that can also function as a repressor element. Thus, the knot regulatory region in D. melanogaster has two apparently functionally redundant blocks of sequences for repression by UBX, both of which are widely separated from activator sequences. This redundancy suggests that the complete evolutionary unit of regulatory control is larger than the minimal experimentally defined control element. The span of regulatory sequences upon which selection acts may, in general, be more expansive and less modular than functional studies of these elements have previously indicated.