Butterfly wing patterns have emerged as exceptional model systems with which to link the developmental and genetic processes that generate morphological variation with the ecological and evolutionary processes that shape this variation in natural populations. Among butterflies, research on species within the genus Heliconius has provided remarkable opportunities to explore how phenotypic diversity is generated within the context of an extraordinary adaptive radiation. Wing pattern diversity among the 48 species and hundreds of intraspecific variants arose within the last 12–14 million years and includes striking pattern convergence between distantly related species, as well as marked pattern divergence between closely related populations and species. Here, we synthesize recent research aimed at gaining a mechanistic understanding of how this variation is generated. This research integrates decades of controlled crossing experiments, and the discovery of major wing patterning genes (optix, aristaless1, WntA and cortex) with recent functional genetic manipulation using CRISPR/Cas9 targeted mutagenesis. The emerging data provides a rich framework with which to explore the repeatability of evolution, particularly within the context of how natural selection acts on divergent gene regulatory networks to generate both highly convergent, as well as highly divergent phenotypes. Overall, the functional data show that the gene regulatory networks underlying pattern variation diverge rapidly in Heliconius; yet these networks retain enough flexibility so that natural selection can drive the evolution of nearly identical patterns from different developmental genetic starting points. Moreover, for the first time this research is starting to illuminate the links between the genetic changes modulating pattern variation and how they influence the larger gene networks that are ultimately responsible for patterning a butterfly wing. There are still large gaps in our understanding, but current research priorities are well laid out and experimental methodologies are in place to address them. The challenge is to synthesize diverse research strategies into a cohesive picture of morphological evolution.