In the water-scarce environment of South Africa, drought-tolerant eucalypt species have the potential to contribute to the timber and biomass resource. Biomass functions are a necessary prerequisite to predict yield and carbon sequestration. In this study preliminary biomass models for Eucalyptus cladocalyx, E. gomphocephala and E. grandis x E. camaldulensis from the dry West Coast of South Africa were developed. The study was based on 33 trees, which were destructively sampled for biomass components (branchwood, stems, bark and foliage). Simultaneous regression equations based on seemingly unrelated regression were fitted to estimate biomass while ensuring additivity. Models were of the classical allometric form, ln(Y) = a+x(1)ln(dbh)+x(2)ln(h), of which the best models explained between 70% and 98% of the variation of the predicted biomass quantities. A general model for the pooled data of all species showed a good fit as well as robust model behaviour. The average biomass proportions of the stemwood, bark, branches and foliage were 60%, 6%, 29% and 5%, respectively.