We report a laboratory-scale study about the suitability of the plasma spraying process for "in situ" repair of cracks in mullite refractories of industrial furnaces. The "design of experiments" approach is used to investigate how the coating porosity and thickness are influenced by six experimental parameters. Arc current, secondary gas (H 2) flow rate, and stand-off distance are the most significant parameters for both responses. Several interaction terms also affect significantly the thickness response. The validity of the model equations is discussed both from a statistical point of view and regarding the physical credibility of the main model terms. Additional experiments confirm that the measured properties lie into the prediction intervals provided by the model. Using a set of parameters optimized for minimal porosity and high thickness (relevant for the crack repair application), coatings with 6% porosity and 1070 μm thickness can be prepared reproducibly. © 2013 ASM International. ; Peer reviewed