We develop a model of induced innovation that applies to medical research. Our model yields three empirical predictions. First, initial death rates and subsequent research effort should be positively correlated. Second, research effort should be associated with more rapid mortality declines. Third, as a byproduct of targeting the most common conditions in the population as a whole, induced innovation leads to growth in mortality disparities between minority and majority groups. Using information on infant deaths in the United States between 1983 and 1998, we find support for all three empirical predictions. (Contains 6 figures, 7 tables and 24 footnotes.)