Noogoora burr (Xanthium occidentale), an annual species, is an important weed of sheep grazing lands in eastern Australia. This paper describes a model of Noogoora bun population dynamics which enables alternative management and control procedures to be simulated. The model is based on a two-phase life cycle with a long-lived overlapping population of seeds and short-lived non- overlapping populations of plants. The model takes into account changes in soil seed reserves due to germination, decay and dispersal. Density dependent processes affecting seed production were incorporated by using Shinozaki's reciprocal yield equation and measure- ments were made of seed losses due to granivory by birds and mice. Simple rainfall records were used to derive the model which was used to predict changes in seed populations in various environments. Theoretical seed populations were close to those actually observed. The model was also used to predict how particular control measures might affect population densities and the types of organisms which might be appropriate for biological control of the weed. Biological control appears to be the only possible means of controlling the large scale infestations of Noogoora burr in semi- arid areas where landholders have abandoned conventional control techniques.