We show that under arguably plausible assumptions regarding the DNA exoneration process, in expectation, the ratio of DNA exoneration rates across races among defendants convicted for the same crime in the same state provides an upper bound on the ratio of wrongful conviction rates across races among these defendants. Our estimates of this statistic reveal that among those sentenced to incarceration for rape in the United States between 1983 and 1997, the wrongful conviction rate among white defendants was less than two-thirds of what it was for black defendants. Our results with respect to murder are inconclusive.