The "Smart Walking" Monte Carlo algorithm is examined. In general, due to a
bias imposed by the interbasin trial move, the algorithm does not satisfy
detailed balance. While it has been shown that it can provide good estimate
s of equilibrium averages for certain potentials, for other potentials the
estimates are poor. A modified version of the algorithm, Smart Darting Mont
e Carlo, which obeys the detailed balance condition, is proposed. Calculati
ons on a one-dimensional model potential, on a Lennard-Jones cluster and on
the alanine dipeptide demonstrate the accuracy and promise of the method f
or deeply quenched systems. (C) 2001 American Institute of Physics.