This paper seeks to examine how federal financial support of academic scientists shapes their collaborative behavior. More specifically, it investigates two questions: (i) are there threshold sizes of federal grants that have effects on collaboration? and (ii) do the threshold sizes differ by discipline? To test and model these questions, this paper uses data collected in a 2004–2005 survey by the Research Valuing Mapping Project (a project based at Georgia Tech and led by Barry Bozeman) of 2,086 academically based research scientists and engineers. The analysis in this paper reaffirms the positive relationship between federal funding and collaboration seen in previous works, but the findings suggest that there is in fact a threshold effect—as the positive relationship has a diminishing influence. The findings of this paper are more limited with respect to how academic disciplines’ threshold levels differ. The results indicate a difference in the threshold levels between disciplines for three of the eight models.