Management practices for seed production of Digitaria eriantha cv. Premier were studied in four experiments at the University of Queensland Farm at Redland Bay, south­ east Queensland. The objectives were to investigate the effect of rate of nitrogen fertilisation, timing of harvest, timing of cleaning cut, timing of tiller emergence, and paclobutrazol application on crop development, seed yield and seed yield components of Premier. In addition, a comparative study was undertaken at the C.S.I.RO. Research Station at Samford, south-east Queensland, from August 1989 to April 1990, to assess the seed yielding potential of cvv. Premier and Advance. In all these experiments, a minimum of three seed harvests per crop was taken and maximum seed yield was used to compare the treatments. Plots were irrigated to avoid water stress. Since the species flowers throughout most of the growing season, two seed crops per year were obtained. Seven levels of N (0 - 300 kg/ha/crop) were tested. The optimal level of N application varied between seasons and years. Maximum seed yield (four to seven times the control treatment) was obtained with 50 and 150 kg/ha/crop in spring 1990 and 1991, respectively. The lack of response to higher levels of N was associated with lodgi.ng, which occurrd at or before anthesis. In genera the effect of N on the autumn crop was inconsistent. In autumn 1990 N application slightly increased seed yield, whereas in autumn 1991 N application tended to reduce seed yield. The low seed yield of the autumn crop may reflect the sudden removal of material from the previous crop, which resulted in poor synchrony of tiller emergence early in the development of the crop. In addition, wet and overcast weather during flowering and seed maturation may have reduced seed set and seed weight per inflorescence. Increased seed yield through N application resulted mainly from increased tiller fertility; also, seed weight per inflorescence was increased in most cases. N application had a negligible effect on seed germination, viability and 1000-seed weight, but increased N content of the seed. In general, maximum seed yield occurred two weeks after peak anthesis. This interval however, varied depending on the local conditions, especially in terms of temperature, rain/all and wind speed. The effect of cleaning cuts made at different times from mid-August to mid­ October was assessed. When the cleaning cut was made before the onset of the reproductive phase (late-September) synchronisation of inflorescence emergence (and thus crop synchronisation) improved, seed weight per inflorescence increased and consequently seed yield increased. In spring 1990, the cleaning cut made late in September increased seed yield by 90% when compared with the control (no cleaning cut), but in the following spring (1991) there was no difference between the cleaning cut treatments made before the end of September (late vegetative phase). On the other hand, cleaning cuts made after the onset of the reproductive phase (mid-October) drastically reduced seed yield. This seed yield reduction is discussed in terms of the removal of reproductive apices, re­ allocation of assimilate to restore the photosynthetic area, and the shaded environment at the base of the plant. In the autumn crop the follow-on effect of spring cleaning cut treatments was negligible. Tillers which emerged within one week of the cleaning cut were the largest and the main contributors to seed yield. This effect was consistent in both autumn and spring crops. The application of N increased the percentage contribution to total seed yield of tillers which emerged 7 - 14 days after the cleaning cut. Overall N increased tiller fertility and tiller dry matter weight, but also reduced the survival of all tiller cohorts; this effect was more evident in the late-emerged tillers and in autumn. Seed yield was increased by 52% with the application of 0.75 kg/ha paclobutrazol at initial inflorescence emergence. This increase was associated mainly with an increased seed weight per inflorescence and a reduction in lodging. Although paclobutrazol tended to increase the number of vegetative tillers, it reduced markedly the number of aerial tillers, which together with the reduction in the stem length would have resulted in more assimilates available to the growing seed. Early applications of paclobutrazol (at cutting or at floral initiation) reduced inflorescence density and consequently reduced seed yield. Effects of the residual spring-application of paclobutrazol were evident in the autumn crop; however, in most cases, differences between seed yield components were not significant. The comparison between cvv. Premier and Advance indicated that both cultivars have similar seed yield potential Nevertheless, they differed in the pathways by which the maximum seed yield was reached. Advance had a high tiller fertility and low botanical fertility, whereas Premier had a low tiller fertility but a high botanical fertility. It was concluded that (i) Application of nitrogen has the potential for increasing seed yield of Premier, (ii) Lodging, which may occur as a consequence of N application, can be successfully controlled with paclobutrazol, and (iii) The critical time for harvest occurred about two weeks after anthesis, which may be controlled appropriately by the timing of the cleaning cut.