We propose that contour nasals come from two principal sources. One source, articulatorily driven, comes from underlying voiced stops, as nasal venting in order to sustain voicing. The other, perceptually driven, comes from underlying nasal consonants, as shielding next to contrastively oral vowels. Although both processes are phonetically well motivated, we argue that the contoured allophones specifically arise in languages in which systemic or phonotactic restrictions allow for easy recoverability of the corresponding underlying segment. Finally, we present a few cases of contour nasals in preconsonantal contexts that seem to be neither venting nor shielding, and suggest that these arise due to place-of-articulation enhancement in clusters. We offer diagnostics for distinguishing nasal venting from shielding and present case studies from South American languages in which understanding such phenomena as enhancement involves analytical commitments to what is contrastive in the language.