Hyperspectral images of galvanized steel plates, each containing a stain of cyclotrimethylenetrinitramine (RDX), were recorded using a commercial long-wave infrared imaging spectrometer. Demonstrations of passive RDX chemical detection at areal dosages between 16 and 90 microg/cm(2) were carried out over practical standoff ranges between 14 and 50 m. Anomaly and target detection algorithms were applied to the images to determine the effect of areal dosage and sensing distance on detection performance for target RDX. The anomaly detection algorithms included principal component analysis, maximum autocorrelation factors, and principal autocorrelation factors. Maximum difference factors and principal difference factors are novel multivariate edge detection techniques that were examined for their utility in detection of the RDX stains in the images. A target detection algorithm based on generalized least squares was applied to the images, as well, to see if the algorithm can identify the compound in the stains on the plates using laboratory reflection spectra of RDX, cyclotetramethylenetetranitramine (HMX), and 2,4,6-trinitrotoluene (TNT) as the target spectra. The algorithm could easily distinguish between the nitroaromatic (TNT) compound and the nitramine (RDX, HMX) compounds, and, though the distinction between RDX and HMX was less clear, the mean weighted residuals identified RDX as the stain on the plate. Improvements that can be made in this detection technique are discussed in detail. As expected, it was found that detection was best for short distances and higher areal dosages. However, the target was easily detected at all distances and areal dosages used in this study.