We employed a computational modeling approach to study the Michaelis complexes of (R)- and (S)-propranolol with serine-acylated Candida antarctica lipase B using four acyl groups: ethanoyl, butanoyl, octanoyl and hexadecanoyl. Our methodology involves sampling Michaelis complex conformations, first through ensemble docking using consensus scoring, and second by molecular dynamics simulations employing a quantum mechanics/molecular mechanics approach. The conformations are then categorized into two classes of near attack conformations, according to the distance of (a) the amino and (b) the hydroxy group of propranolol to the catalytic residues. The relative populations of these two classes of conformations was found to be consistent with the experimentally-observed exclusive chemoselectivity toward O-acylation with ethanoyl. Furthermore, we predict that increasing the length of the hydrocarbon chain of the acyl group will cause O-acylation to be unfavorable.