Sociologists of religion have often connected secularization to science, but have rarely examined the role of religion in the lives of scientists or how the sciences have changed religiously over time. Here we address this shortcoming by comparing religiosity between two samples of elite academic natural and social scientists, one in 1969 and one in 2005. Findings show an overall decline in religiosity among university scientists as well as a change in their religious composition. Attendance rates were lower for social scientists in 1969 compared to natural scientists, but in 2005 growing parity in attendance occurred between the two fields. Findings also show a decline in the proportion of Protestant scientists and a growth in Catholic scientists. Demographic factors associated with religiosity in the general population, with the exception of age and having children, had no impact among elite academic scientists. Overall, findings challenge and revise older studies on the role of religion in the lives of scientists. Specific results are connected to theories that pose science as a master identity, which may be mitigated by some institutionalized aspects of religion. They also lead to new directions in the sociology of religion that take seriously the role of religion in the lives of elites and connect societal religious changes to differences between institutional spheres, particularly those, such as the academy, that play a leadership role in society. (Contains 3 tables and 16 notes.)