One of the reasons for the ongoing heated debates on the importance and reliability of morphological characters in phylogenetic reconstruction is the lack of understanding of how morphology evolves. One way to examine morphological evolution is to reconstruct evolutionary transformations of characters on existing phylogenies using maximum parsimony, maximum likelihood, and Bayesian approaches. In this study, I reconstructed the evolution of larval, pupal, and adult character suites of metalmark moths (Lepidoptera: Choreutidae) on a molecular phylogeny inferred using one mitochondrial (cytochrome oxydase I) and two nuclear (elongation factor 1-alpha and wingless) genes. This assessment shows widespread homoplasy in seemingly complex morphological character suites that were thought to represent synapomorphies at the outset of the study. For example, one such character suite is antennal scalation: in most choreutids antennae are covered with short scales that form alternating white and black rings for the entire length of the antenna; others have a covering of long scales restricted to the basal two thirds of the antenna and a longitudinal white stripe near the tip. Taxa with strikingly similar antennae are inferred to be distantly related. Repeated losses, parallelism, and convergence are possible explanations for homoplasy; however, the underlying reasons, such as the evidence for natural selection, are unknown. These findings are significant because they underscore the challenges of using morphological characters in phylogenetic reconstruction and in revising classifications.