Why are animal signals so diverse? This question has received intense scrutiny, and both natural and sexual selection have been implicated as important drivers of signal diversity. Under the framework of sensory drive natural selection is expected to favor signals and receptors that maximize signal detection. A corollary of this is that the genes underlying both signal production and reception should show evidence of selection as they evolve to maximize detection. Fireflies are an excellent system in which to study signal and receptor evolution because the primary genes underlying each are known. Fireflies vary in their signal color, but color is not important for female choice, implying that natural selection for detection is the primary driver of color evolution. Sequence variation in luciferase, the enzyme that catalyzes the light reaction, is thought to cause the color differences between species. Likewise, sequence variation in opsins, the protein components of visual pigments, may underlie the variation in visual sensitivity, which has been shown to match signal color, across species. Here we tested the hypothesis that light color is due to selection acting on luciferase and the resulting predictions that (1) sequence variation in luciferase correlates with observed variation in signal color, (2) sequence variation in opsins correlates with observed variation in signal color (and inferred visual sensitivity), and (3) signatures of selection are evident at these two loci. To do this, we sequenced the luciferase and opsin genes of 192 individuals from 12 populations of the North American firefly, Photinus pyralis. We also sequenced the cytochrome oxidase I (COI) locus and genotyped individuals at 716 single nucleotide polymorphisms (SNPs) using triple digest restriction site associated DNA sequencing (3RAD). We tested for selection while accounting for population structure by comparing differentiation (Fst) at SNPs in luciferase and opsins to the distribution of Fst for the genome-wide 3RAD SNPs. We found no variation at the amino acid level in either luciferase or opsins despite the population differences in signal color. However, silent variation in luciferase, but not opsins, shows high levels of differentiation among populations, strongly suggesting that selection is acting at this locus. The absence of protein variation rejects the paradigm that variation in light color is solely due to variation in luciferase sequence in P. pyralis. Instead, natural selection appears to target regulatory or other non-coding variants. If natural selection targets light emission color, our work suggests that either non-coding luciferase variation directly alters emission color through an unknown mechanism, or changes at another gene are correlated with changes in non-coding variants in luciferase.