There is currently a global increase in Eucalyptus pulpwood plantations. Harvesting systems traditionally utilised in the northern hemisphere are being used in Eucalyptus pulpwood plantations worldwide. However, the small tree size and complexity of debarking Eucalyptus have provided harvesting with productivity and cost challenges not previously experienced in northern-hemisphere conditions. Much research has been invested in these two harvesting methods in northern-hemisphere species and conditions. There is little research available on mechanised processing-machine productivity and costs in Eucalyptus. This investigation aimed to quantify the effect that tree and bundle size has on the productivity of different processing machines in Eucalyptus plantation pulpwood. This was done through regression analysis, whereby productivity models that included tree size and bundle size were constructed. The research also aimed to determine whether or not the multi-stem systems were more cost-effective in smaller tree sizes. The research investigated five mechanised harvesting options that forestry managers could use in Eucalyptus pulpwood plantations. These systems consisted of one CTL system, one full-tree system with single-stem processing and three full-tree systems with multi-stem processing. The CTL system used a harvester to process the trees into logs and to extract them. The full-tree system with single-stem processing used a dangle-head processor (DHP) to process trees into logs. The first full-tree system with multi-stem processing used a chain-flail debrancher debarker (CFDD) to produce debarked and debranched tree lengths, which were slashed into logs. The remaining full-tree, multi-stem systems both produced chips. The first used a chain-flail debrancher debarker chipper (CFDDC) and the second, a CFDD feeding into a stand-alone disc chipper (CFDD&C). The productivity data, measured as m3 per productive machine hour (PMH), was then statistically analysed using regression techniques. Productivity equations were formulated, considering tree size and bundle size, as well as the quadratic functions of these two variables and the interaction between them. Bundle size was only applicable to the multi-stem processing machines. The productivity equations successfully predicted processing-machine productivity, using tree size and bundle size as input variables. Apart from the 0.075 m3 tree size class, the CFDD had the highest overall productivity. The costs of the five systems were then calculated for different tree sizes. No single system was more cost-effective than the others across all tree sizes. In 0.075 m3 trees, the CFDDC system proved the most cost-effective. All systems evidenced high costs in the 0.075 m3 trees, ranging between $19.43 per m3 for the CFDDC system to $28.84 for the harvester system. In 0.40 m3 trees, the cost differences between systems were lower, ranging from $6.91 per m3 for the DHP system to $11.84 per m3 for the CFDD&C. This study confirms that the CTL system was very expensive to operate in the small tree sizes (0.075 m3). There is a cross-over point at 0.25 m3 per tree, where the CTL system costs become lower than those of the full-tree system. At the 0.40 m3 tree size, the full-tree system is slightly more expensive than the CTL system. Copyright / Dissertation (MSc)--University of Pretoria, 2012. / Plant Production and Soil Science / unrestricted