Chiral a-and b-hydroxy amides are useful building blocks for the synthesis of biologically active compounds.[1] The synthesis of these intermediates in both a regio-and stereoselective manner, however, is difficult. There are only a few methods for the synthesis of such chiral units, and their substrate scope and selectivity remain unsatisfactory.[2] The regioselective epoxide-opening reaction of optically active a,b-epoxy amides is one of the most attractive approaches to this problem. We [3] and Aggarwal's group [4] recently succeeded in developing efficient strategies to obtain a,b-epoxy amides in a highly enantioselective manner. There are no reports, however, on regioselective epoxide-opening reactions of a,b-epoxy amides, [3,5] in contrast to the success with a,b-epoxy ketones.[6] We report herein a new synthesis of nearly enantiomerically pure a-and b-hydroxy amides with high substrate generality, which consists of a novel highly regioselective epoxide opening of both b-alkyl-and b-arylsubstituted a,b-epoxy amides (Scheme 1). An efficient enantioselective synthesis of (R)-fluoxetine, using the newly developed method, is also described.To realize the highly regioselective epoxide-opening reactions of a,b-epoxy amides, it is important to control the relative reactivity of the a-and b-positions, which depends on the b-substituent. Thus, we discuss the reactions of the b-arylsubstituted amide (paths A and B) and of the b-alkylsubstituted amide (paths C and D).We recently described the highly enantio-and regioselective synthesis of b-aryl a-hydroxy amides using a one-pot, tandem process consisting of catalytic asymmetric epoxidation and a Pd-catalyzed epoxide opening (path A). The selectivity was based on the higher reactivity of the b-position (benzyl position) over that of the a-position.[3] The higher reactivity of the b-position, however, make it difficult to obtain b-hydroxy amides through cleavage of the C a ÀO bond (path B). Indeed, the general conditions for selective cleavage of the C a ÀO bond in a,b-epoxy ketones, such as SmI 2 and [Cp 2 TiCl 2 ]/Zn, [6,7] gave unsatisfactory results (trace amounts) with a,b-unsaturated and -saturated amides as the major products. To overcome this difficulty, we examined a so-called intramolecular hydride transfer using Red-Al (sodium bis(2-methoxyethoxy)aluminum hydride), [1c, 8] which might react with NÀH first to produce a NÀAl species; the remaining hydride attacks the a-position of the epoxy amide (see Figure 1). As we expected, the reduction of 2 a with Red-Al gave b-hydroxy amide 5 a [7] as the major product in moderate yield (Table 1, entry 1). This result prompted us to optimize the reaction conditions.To gain insight into the reaction mechanism, especially for the counterion effects, calculations were performed by means of the hybrid density functional method (B3LYP [9] ) using a 6-31G(d) basis set. As shown in Figure 1, the coordination of a sodium ion to the epoxide and carbonyl oxygen atoms should weaken the C b ÀO bond (D b = 0.0315 ) more effectively ...