How are university faculty members in STEM disciplines motivated to conduct research, and how does motivation predict their success? The current study assessed how multiple types of self-determined motivation predict research productivity in a sample of 651 faculty from 10 US institutions. Using structural equation modeling, the basic psychological needs of autonomy and competence predicted autonomous motivation (enjoyment, value) that, in turn, was the strongest predictor of self-reported research productivity. Using negative binomial regression, autonomous motivation was the strongest predictor of faculty publications and citations, with a one-standard deviation increase in autonomous motivation (approximately a half response option on a 1-5 Likert scale) corresponding to an 11.63% increase in publications and a 22.57% increase in citations over a three-year period. Occupational and social-environmental background variables (e.g., research percentage on contract, career age, balance, collegiality), as well as controlled motivation (guilt, rewards), had comparatively limited predictive effects. These results are of relevance to higher education institutions aiming to support scholarly productivity in STEM faculty in identifying specific beneficial and detrimental aspects of faculty motivation that contribute to measurable gains in research activity.