Treatment of a 1:1:1 mixture of enolate 5:amine 6:lithium amide 7 with BF3.OEt(2) affords the naproxen amide 4a with an enantiomer excess of 77% (>90% yield). The result is attributed to Lewis acid-induced internal proton return (ipr) in a mixed aggregate containing the enolate and the chiral amine. Use of proline-derived diamines 9 and 10 in place of 6 affords 4a with 56-66% ee, but monoamines are relatively ineffective. Similar ipr conditions can be used to deracemize the oxazolidine 13 (50-60% ee), the lactam 14 (50% ee), and the cyclohexenyl propionamide 15 (62% ee). However, disappointing results were obtained with 6 and several esters and the lactone 18. Lactone 18 was deracemized with the diamine 24 (70% ee) under ipr conditions, but simple acyclic esters gave marginal ee values with 24 (BF3.OEt(2) quench). Better results were obtained with methyl N-benzoylalaninate 16 (73% ee). In the latter case, the dianion was generated and ipr was induced by the sequential addition of 24 and BF3.OEt(2) as before. In the case of amide 4a, H-1 NMR evidence shows that much of the proton transfer is complete before the addition of BF3.OEt(2) to the solution of 24 and enolate 5. Thus, 5 is quenched by direct proton transfer, not by ipr, when 24 is used as the chiral amine. The proton transfer pathway can be correlated qualitatively with pK(aDMSO) values. Thus, 24 was found to have a pK(aDMSO) = 27.7 while the value for 4a is ca. 31. The relative acidity in THF is assumed to be similar, and 24 can protonate 5 directly but not the lactone enolate 19 (18 : pK(a(DMSO)) = 20.1). Direct proton transfer does not occur with 6 (estimated pK(aDMSO) = ca. 34-35) with any of the enolates studied, and activation for ipr by BF3.OEt(2) is necessary to activate the N-H bond. In several examples, protic acid-induced ipr was also explored. In all cases, this gave lower ee values than the BF3.OEt(2) method.