Abstract We estimate the first econometric model of the national civilian firearms market in the United States (1946–2016), where per capita firearms‐related harm is exceptionally high. Solving simultaneous equation models instrumented by natural disasters and steel prices, and employing unique firearms prices and quantities data, we find this market operates normally, except that firearms stocks may generate some new market demand in a positive feedback loop. Save for the Federal Assault Weapons Ban (1994–2004), federal firearms legislation does not influence firearms sales. We find that violent crime, including homicide and mass shootings, boosts domestic sales.