Plant water stress has been extensively studied using hyperspectral visible- and near-infrared systems. Thermal imaging and the recent availability of widely tunable infrared quantum cascade laser (QCL) allow us to propose an active hyperspectral imaging system operating in the mid-infrared (MIR) band, where the system output consists of a series of narrowband subimages arranged across the reflectance spectrum of the sample, forming a hypercube data acquired by “staring” acquisition technique. To evaluate more precisely the capabilities of the active hyperspectral imaging, we propose a system composed of four powerful tunable QCL covering the 3.9- to 4.7-μm and 7.5- to 11-μm wavelengths ranges. Two cameras are used for detection: an InSb cooled camera ranging from 3 to 5  μm and a bolometer from 7.5 to 13  μm range. This system is validated by applying to growing plants for early water stress detection. Finally, we present and discuss results using partial least squares discriminant analysis classification technique to characterize water status of different plants, separated in two classes: control subjects were maintained at 80% of the amount of water to soil saturation ratio and stressed subjects at 20%. Initial discrimination results have shown the efficiency of the proposed system.