The effectiveness of a new algorithm, parallel tempering, is studied f
or numerical simulations of biological molecules. These molecules suff
er from a rough energy landscape. The resulting slowing down in numeri
cal simulations is overcome by the new method. This is demonstrated by
performing simulations with high statistics for one of the simplest p
eptides, Met-enkephalin. The numerical effectiveness of the new techni
que was found to be much better than traditional methods and is compar
able to sophisticated methods like generalized ensemble techniques. (C
) 1997 Elsevier Science B.V.