Striking trait polymorphisms are worthy of study in natural populations because they can often shed light on processes of phenotypic divergence and specialization, adaptive evolution, and (in some cases) the early stages of speciation. We examined patterns of genetic variation within and between populations of mormyrid fishes that are morphologically cryptic in sympatry but produce alternate types of electric organ discharge (EOD). Other species in a large group containing a clade of these morphologically cryptic EOD types produce stereotyped, species-typical EOD waveforms thought to function in mate recognition. First, for six populations from Gabon's Brienomyrus species flock, we confirm that forms of electric fish that exhibit distinctive morphologies and unique EOD waveforms (i.e., good reference species) are reproductively isolated from coexisting congeners. These sympatric species deviate from genetic panmixia across five microsatellite loci. Given this result, we examined three focal pairs of syntopic and morphologically cryptic EOD waveform types that are notable exceptions to the pattern of robust genetic partitioning among unique waveform classes within assemblages. These exceptional pairs constitute a monophyletic group within the Brienomyrus flock known as the magnostipes complex. One member of each pair (type I) produces a head-negative EOD, while the other member (either type II or type III, depending on location) produces a longer duration EOD differing in waveform from type I. We show that signal development in these pairs begins with juveniles of all magnostipes-complex morphs emitting head-positive EODs resembling those of type II adults. Divergence of EOD waveforms occurs with growth such that there are two discrete and fixed signal types in morphologically indistinguishable adults at each of several localities. Strong microsatellite partitioning between allopatric samples of any of these morphologically cryptic signal types suggests that geographically isolated populations are genetically decoupled from one another. By contrast, sympatric morphs appear genetically identical across microsatellite loci in Mouvanga Creek and the Okano River and only very weakly diverged, if at all, in the Ivindo River. Our results for the magnostipes complex fail to detect species boundaries between the focal morphs and are, instead, fully consistent with the existence of relatively stable signal dimorphisms at each of several different localities. No mechanism for the maintenance of this electrical polymorphism is suggested by the known natural history of the magnostipes complex. Despite a lack of evidence for genetic differentiation, the possibility of incipient sympatric speciation between morphs (especially type I and type II within the Ivindo River) merits further testing due to behavioral and neurobiological lines of evidence implying a general role for stereotyped EOD waveforms in species recognition. We discuss alternative hypotheses concerning the origins, stability, and evolutionary significance of these intriguing electrical morphs in light of geographical patterns of population structure and signal variation.