Studies have been completed into the spectral response of alkaline and other ultramafic rocks from arid and semi-arid regions. To date these rocks have not been investigated spectrally in a systematic fashion using the latest field and airborne imaging spectrometers. The objective was to determine how spectral geological techniques could be used to locate these rock types when they are exposed, weathered and reduced to residual soil. The data used in this study were spectra obtained from field spectrometers (PIMA and GER IRIS MkIV) and airborne scanners (GEOSCAN MkII, GER IS and HyMap). These data were gathered at four sites within Australia. The data processing software packages used for the analyses were commercially available image processing systems (is S'600and ENVI) and a modified version of PIMAVIEW for processing spectra. Spectra were measured, in the field and the laboratory, of alkaline and other ultramafic rocks to determine if they had diagnostic spectral absorption features. These studies demonstrated that there are diagnostic spectral absorption features common to alkaline and other ultramafic rocks including dunite, peridotite and serpentinite. The diagnostic spectrum in the SWIR2 region (2000nm to 2500nm) has absorption features located near 2300nm and 2380nm and results from Mg-OH bearing minerals including serpentine, talc and phlogopite. The VNIR spectra of these rocks also have distinct absorption features, but since these are caused by minerals that are common to a variety of rocks and soils, they have not been considered in this study. When weathered under arid and semi-arid conditions, ultramafic rocks break down into smectite clays, primarily saponite. Saponite has a similar spectral signature to the primary Mg-OH bearing minerals. Further weathering and removal of magnesium results in saponite altering to kaolinite and then to opaline silica. These minerals may occur in residual soils derived from ultramafic rocks. Spectral investigation of mixtures of saponite and other minerals showed that there are linear changes the absorption features of spectra, depth, wavelength and shape, as the proportion of saponite to other minerals varies. The ability to identify ultramafic rocks from the distribution of their diagnostic spectra was confirmed by analysing data obtained from surface samples and airborne scanners. The coincidence of results obtained from field studies and airborne scanner data signifies that data acquired remotely are as useful as spectra obtained in the field, for mapping the extent of ultramafic rocks. Present and past imaging systems have been investigated to determine the specifications required to suit this application. These investigations included sampling the spectra of ultramafic and background rocks with the band pass characteristics of various instruments. The signal-to-noise ratio that is required of scanner data to ensure it is useable was also studied. These studies indicated a specification for data of at least eight and preferably, thirty two channels in the SWIR2 (between 2000nm and 2500nm) obtained with a signalto-noise ratio in excess of 200: 1, ideally 400: 1 at 2200nm. Scanners producing data with these specifications can be used to locate ultramafic rocks from their spectral signature, whether exposed or covered by residual soils in a variety of geological environments. The success of applying these techniques will however depend on the spectral contrast between ultramafic rocks and their backgrounds. Rocks that produce Mg-OH bearing minerals or have near 2300nm absorption can result in areas being mistakenly identified as containing alkaline and other ultramafic rocks. However, in this study it was demonstrated that carbonate soils and rocks that have a near 2300nm absorption feature can be spectrally distinguished from alkaline ultramafic derived saponite using HyMap scanner data. In South Australia dolomite appears to have altered to saponite at surface but kimberlites in the area can still be detected by processing HyMap scanner data with spectral as opposed to conventional image processing techniques.