An area and a point-based technique were used together at each of the same 75 sampling sites (stands), on a Sourish-Mixed Bushveld farm, to collect data for the classification and mapping of the vegetation. Both sets of data were synthesized using the same computer program package and the efficacy of the resulting classifications as well as the efficiency of the two field sampling techniques was compared. Following this, a continuous 7 752 point (1 m apart) transect was carried out, traversing the farm, in order to determine the optimum scales at which to sample Sourish-Mixed Bushveld so as to increase classification efficacy and improve community boundary recognition. The results indicated that (1) the arbitrarily chosen sampling scale of 1:8 000 was too large for "farm-scale" studies; (2) the area-based method proved to be satisfactory in that the classification and vegetation map produced with this method were verified spatially and environmentally; (3) the point-based method was deficient as a classificatory and mapping tool at large scales, since too few species were recorded with this method to make any sense of the classification and mapping of the vegetation was not possible; (4) less time per species was spent using the area-based method but because more species per stand were recorded with this method, the point-based method was quicker per stand; (5) the area-based method was easier to use in dense vegetation and irregular terrain; and (6) the optimum sampling scales for Sourish-Mixed Bushveld, as indicated by the synthesis of the continuous transect data, are about 1:12 000, 1:50 000 and 1:250 000. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 1995.