Classical-mechanics-based methods are the major theoretical tool in modern biomolecular simulations.[1] However, their application is limited to classical systems that do not exhibit strong quantum mechanical (QM) effects. Among these, intermolecular charge transfer (CT) in halide-water clusters has been linked to the red-shifting of hydrogen-bonded OH stretching vibrations, [2] which has been confirmed by X-ray absorption spectroscopy.[3] Since biological macromolecules are ionized at physiological conditions, a proper understanding of the CT effect is of utmost importance to understand biomolecular systems in solution. Early studies on small-molecule clusters [4][5][6][7] dismissed CT as a negligible effect. However, Merz and co-workers, [8,9] based on a semiempirical PM3 [10] calculation, predicted about two electron units of CT from the neutrally charged, 69-residue, Cold-Shock protein A to water. Additional evidence came from an HF/6-31G* study of Komeiji et al. [11] who predicted-on another neutrally charged 76-residue protein, namely, ubiquitin-0.8 electron units of CT to water. Although these studies clearly showed the significance of CT, they were based on "a posteriori" single-point QM calculations on classical trajectories. Since semiempirical methods are likely to be more sensitive to the choice of non-native geometry than ab initio methods (due to their parametric nature), this casts some doubts about the quantitative accuracy of those results. Moreover, an eventual QM minimization of the classical trajectory would map it onto local minima on the QM energy hypersurface, thus rendering this approach inadequate to study the dynamic processes.To take a further step in the application of more accurate methods to study CT effects, we performed 20 ps of full QM molecular dynamics (QM MD) of ubiquitin in a water droplet (explicit solvent), using semiempirical AM1, [12] RM1, [13] PM3, [10] and PM5 [14] Hamiltonians. For the sake of comparison with earlier approaches, [8,9,11] single-point QM calculations over molecular mechanics (MM) MD trajectories were also performed. An intermediate approach consisting of QM energy minimization of 1000 MM MD snapshots was also considered to assess the necessity of protein relaxation in QM energy evaluation of classical trajectories.The results of semiempirical single-point calculations of ubiquitin in water based on 1000 MM MD snapshots are displayed in the first row of Table 1. PM3 calculations predicted the largest CT of À2.0 units, very similar to the data of Merz and co-workers.[8] Such coincidence may stem from the neutrality of both proteins, and their similar number of total and charged amino acids. Our study also showed a tendency of PM3 to an overestimation of CT.[15] Although the AM1 value of À1.2 was close to the value reported by Komeiji [11] (i.e., À0.8), AM1 showed other drawbacks, as will be discussed later.The QM MD revealed an unphysical tendency of AM1 and RM1 to dissociate the OÀH bond of water which was observed during the first ps of the AM1 simulatio...