This experiment evaluated the relationship between carrying capacity and veld condition. Goat browsing days were used as a measure of shrub productivity and cattle grazing days as a measure of the sward. Twigs on two shrub species were monitored to quantify utilization and recovery of the woody layer and a pasture disc meter was used to monitor grazing of the grass layer. Dense bushveld (high ecological status) had higher overall carrying capacity than open bushveld (low ecological status). Carrying capacity changed considerably depending on annual rainfall and this was complicated because the more degraded the veld the greater the variation. Maximum long‐term grass production occurred at intermediate tree densities. This was, however, extremely unreliable between years and a permanent grazer enterprise cannot be recommended. For total forage production and constant fodder supply, dense bushveld is optimum. Optimum, however, changes from season to season and with operator objectives. In wet seasons it will pay to have veld in the lower range of ecological status (i.e. open bushveld) whereas in other seasons veld with high ecological status (dense bushveld) will be more productive. Farmers with only cattle may aim to have veld in the mid‐range of ecological status but this conflicts with farmers who have browsers (e.g. goats and kudu) and with the conservation ideal, because bushveld, once thinned, does not regenerate, thereby limiting future management alternatives.