Wildlife monitoring is an important conservation tool, but in the savannah regions of Africa, cash-strapped and capacity-limited authorities rank it low on their priority list. To try to reduce the time, effort, and financial costs of monitoring large mammals, I examine a 20-year dataset of herbivore records taken from vehicle transects carried out in Katavi National Park, western Tanzania. I find that: (i) population trends obtained from ground transects are similar to those obtained from aerial surveys conducted over a wider area; (ii) the frequency of vehicle surveys driven per year or (iii) across years can be reduced without losing substantial information; (iv) it is inadvisable to stint on numbers of transects driven; and (v) trends in populations of single species do not represent those of others. These findings are encouraging because they indicate that managers can obtain relatively accurate information about herbivore population trends through infrequent and therefore more cost-effective monitoring.