In the field of remote sensing, hyperspectral imagery has recently begun to be exploited for its superior spectral resolution. However, hyperspectral imagery contains redundant information, and therefore band reduction techniques have been developed to effectively compress the information or variance into fewer bands to simplify and expedite exploitation. Similar noise reduction transforms have been used quite extensively as well. Specifically, principal components analysis (PCA) and noise adjusted principal component (NAPC) or maximum noise fraction (MNF) transforms have become standard dimension and noise reduction techniques for hyperspectral data as applied to target detection. These transforms have been commonly used with the assumption, with little or no basis in factual evidence, that there is no adverse effect on detection performance. Therefore, the objective of this study was to test the hypothesis that dimensionality and noise reduction transforms on hyperspectral data have no effect as applied to target detection algorithms. The spectral angle mapper (SAM) and orthogonal subspace projection (OSP) detection algorithms along with two different targets from a HYDICE scene were studied in this research. The overall conclusion made from the results of the dimension reduction experiment was that the detection statistics improved or remained relatively constant till a significant number of dimensions were removed, and in some cases it increased the separation between targets and background classes. The conclusion made from the noise reduction experiment was that the SAM angle improved (i.e. better detection) or remained constant, and target‐background separation increased. Noise reduction using the OSP detection resulted in a statistic that remained constant or deteriorated till a significant number of bands were zeroed. For both experiments once a significant number of dimensions were removed or zeroed the target detection performance drastically deteriorated.