Evolutionary Developmental Biology aims for a mechanistic understanding of phenotypic diversity, and present knowledge is largely based on gene expression and interaction patterns from a small number of well-known model organisms. However, our understanding of biological diversification depends on our ability to pinpoint the causes of natural variation at a micro-evolutionary level, and therefore requires the isolation of genetic and developmental variation in a controlled genetic background. The colour patterns of Heliconius butterflies (Nymphalidae: Heliconiinae) provide a rich suite of naturally occurring variants with striking phenotypic diversity and multiple taxonomic levels of variation. Diversification in the genus is well known for its dramatic colour-pattern divergence between races or closely related species, and for Müllerian mimicry convergence between distantly related species, providing a unique system to study the development basis of colour-pattern evolution. A long history of genetic studies has showed that pattern variation is based on allelic combinations at a surprisingly small number of loci, and recent developmental evidence suggests that pattern development in Heliconius is different from the eyespot determination of other butterflies. Fine-scale genetic mapping studies have shown that a shared toolkit of genes is used to produce both convergent and divergent phenotypes. These exciting results and the development of new genomic resources make Heliconius a very promising evo-devo model for the study of adaptive change.