Even though Daniel Ellsberg’s 1961 article “Risk, ambiguity and the Savage axioms” is well-known and increasingly quoted in current decision theory, introducing the counterexample to Bayesian decision-making that got the normative value of Savage’s theory into trouble, its philosophical background remains totally unknown. This paper examines Ellsberg’s motivations in presenting his critique first to his fellow decision theorists at Harvard and RAND in the late 1950s and it goes into his reasons for giving a philosophical justification and defence of the paradox in his doctoral thesis of 1962. By concentrating mainly on Ellsberg’s all-encompassing analysis of decision-making in his thesis, the paper shows that a number of relevant issues connected to the paradox can be thrown light on. These range from its historical background to the way to test the normative value of decision theory through experiments, and a taxonomy of decision rules based on alternative probabilistic set-ups. Crucially, the paper argues that Ellsberg subscribed to a generalised version of the Bayesian approach, one that informs the developments of the multiple prior approach in current decision theory, but finds its origins in Keynes’s Treatise on Probability.