The theory of subjective expected lexicographic utility brings together two classicaldevelopments in expected utility theory. The first is Hausner''s theory of expected lexico-graphicutility in decision under risk. The second is a lottery-based theory of subjectiveexpected utility in decision under uncertainty that was first axiomatized by Anscombe andAumann. Our synthesis of the two produces representations of preference in decision underuncertainty in which utilities are finite-dimensional real vectors ordered lexicographicallyand subjective probabilities are real matrices. Axiomatizations of subjective expected lexico-graphicutility are described for finite and infinite sets of states. Procedures for assessingvector utilities and matrix probabilities are outlined.