The management and sustainable use of Central Australian rangelands for livestock production and conservation requires improved knowledge of the temporal and spatial distribution of primary production in this region. To provide such information, this thesis investigated methods that could rapidly and efficiently estimate regional herbage biomass production in these arid landscapes. Two different approaches were examined, using (1) ground-based or (2) satellite-based data sources. Soil moisture and herbage growth data were collected over several growth seasons and five landscape types in Central Australia, and the data used to develop a model of soil moisture balance and herbage production for the region. The model has few parameters and only requires inputs of rainfall and potential evaporation to predict daily soil moisture and plant growth. Moisture loss in the 0-500 mm soil profile was modelled using a negative exponential function that depends on available soil moisture and is driven by potential evaporation. The growth of herbage, whilst soil moisture is above wilting point, is a linear function of actual evapotranspiration, with the decay of plant material represented by a logistic curve through time. Soil moisture, herbage biomass and species composition assessments made at hectare and square kilometre scales at four locations within Central Australia were examined to determine if a small sample area could be used to accurately describe the soil and plant conditions at a landscape scale. Moisture levels of the 0-200 and 0-500 mm soil profiles from nine samples were analysed for the beginning and conclusion of a growth season, whilst herbage biomass and species composition from 50 samples were compared at the end of the growth season. Results suggest that mean soil moisture levels determined in a 1 ha area are comparable with mean values in the surrounding 1 km2 area. Herbage biomass and species richness for a square kilometre can be assessed at a hectare site for some landscape types, but a larger sampling area (> 1 ha) is recommended for most rangeland assessments. Satellite data (NOAA-11) were examined for their potential application in assessing primary productivity in Central Australia. Several image correction techniques were tested to minimise the adverse effects of atmospheric contamination and illumination. Two measures of atmospheric moisture: (1) radiosonde data and (2) temperature differences between bands 4 and 5 of the NOAA satellite (split-window) were used to explain variations in NOAA-11 normalised difference vegetation index (NDVI) on inert desert sites. The splitwindow approach provided the best single factor relationship (r2=0.63) and, when combined with scattering angle (illumination) effects, up to 81% of the variation in NDVI data could be explained. Field measurements of herbage biomass were correlated with four growth indices derived from NOAA-11 NDVI data. The influence of preflight and sensor degradation calibrations of Bands 1 and 2, and atmospheric correction techniques were also tested. Correlations between temporal sums of NDVI and herbage biomass data were relatively poor (r2<0.42) and unsuitable for herbage assessment in Central Australia. However, correlations between atmospherically corrected and background-adjusted maximum NDVI data and observed herbage biomass were strong (r2=0.91), that will allow primary production in the arid rangelands of Central Australia to be assessed rapidly and efficiently using remotely-sensed information.