The accuracy of game counts on an eastern Transvaal Lowveld game ranch was examined. A drive count was used to produce control estimates of actual population sizes. A helicopter count was more accurate than one from a fixed-wing aircraft. Twenty-four hour waterhole counts were relatively inaccurate. Line transect counts were relatively accurate when a fixed strip width equal to the mean visibility index of the vegetation was used, and when the population estimate was based on the perpendicular sighting distance.