Many developmental processes are preserved evolutionarily to produce similar if not identical phenotypes across taxa. An unanswered question in evolutionary biology is whether the genes controlling developmentally regulated traits under stabilizing selection to preserve phenotypes are likewise functionally static; or conversely whether functional divergence of individual genes and/or network interactions are inevitable. In this thesis, I address this void in knowledge by investigating the evolutionary dynamics and timescale for the functional divergence of genes in a conserved gene network. First, I ask if sequence divergence in coding and noncoding regions of the gap gene giant (gt) have fitness consequences by replacing the endogenous giant locus in D. melanogaster with transgenic orthologs from six Drosophila species – D. melanogaster, D. yakuba, D. san- tomea, D. erecta, D. pseudoobscura and D. virilis – representing a range of phylogenetic distances spanning the genus Drosophila. By swapping the whole locus —coding and non- coding region of the locus— I document a continuous pace of functional evolution across the giant locus and species-specific coding-noncoding interactions (Chapter 2). Second, I confirmed that the coding region of D. melanogaster giant is responsible for causing embryonic inviability in D. melanogaster/D. santomea hybrids. Further experiments identified the involvement of a second gap gene, tailless, in hybrid inviability. Both giant and tailless from the two species are not functional equivalents. These findings provide additional evidence for a rapid pace of functional divergence of essential gap genes to the extent that they may contribute to hybrid inviability (Chapter 3). Lastly, I investigate aspects of gene expression of the gap gene network in Drosophila. Consistent with the functional evolution of giant, I confirm that the gap gene network outputs are conserved across species, as others have found, but show that this is not true when D. virilis giant is expressed in D. melanogaster, a non-native genetic background. In these flies, outputs of the gap gene network both differ from wildtype and are de-canalized. These findings refute the hypothesis that genes in this essential regulatory network with conserved outputs are themselves functionally static (Chapter 4). To the contrary, conserved outputs require native genetic backgrounds to assure phenotypic constancy, which highlights the importance of co-evolution of genes in the network in maintaining this constancy.