Mineral flotation is commonly used to separate valuable minerals from ore. Effective monitoring and control of a mineral flotation process necessitates accurate and rapid on-line analysis of the elements or minerals in the slurry flows. Recent development of lightweight and low-cost visible and near-infrared (VNIR) imaging spectrographs has provided new interesting possibilities also in mineral froth and slurry analysis. The combination of VNIR reflectance spectroscopy with the traditional X-ray fluorescence (XRF) analysis for improved measurement of the mineral slurry contents is discussed in this article. A sensor fusion approach based on a recursive partial least squares algorithm is used to combine the information from the sparse and asynchronous XRF samples with the high-frequency VNIR analysis. The structure and installation of a prototype measurement device is presented, and results from copper and zinc (chalcopyrite and sphalerite) slurries in a real process environment are discussed. The main improvement achieved by the sensor fusion of the reflectance spectral analysis and the XRF is the significantly reduced sampling interval of the elemental analysis. (C) 2010 Elsevier Ltd. All rights reserved.