The canonical problem of the one-dimensional, pulsating, overdriven detonat
ion wave has been studied for over 30 years, not only for its phenomenologi
cal relation to the evolution of multidimensional detonation instabilities,
but also to provide a robust, reactive, high-speed flowfield with which to
test numerical schemes. The present study examines this flowfield using hi
gh-order, essentially non-oscillatory schemes, systematically varying the l
evel of resolution of the reaction zone, the size and retention of informat
ion in the computational domain, the initial conditions, and the order of t
he scheme. It is found that there can be profound differences in peak press
ures as well as in the period of oscillation, not only for cases in which t
he reaction front is under-resolved, but for cases in which the computation
is corrupted due to a too-small computational domain. Methods for estimati
ng the required size of the computational domain to reduce costs while avoi
ding erroneous solutions are proposed and tested.