The number of parameters and noise factors that influence the plasma spraying process is huge, with over 200 known. Today, only a few parameters such as gas flow rate, current, voltage, spraying distance and substrate roughness can be controlled. In recent years, several particle diagnostic systems have been developed, which give the chance to control processes much better than at present. These techniques are now close to being introduced in industrial applications. In addition to the in-flight particle properties, the surface temperature of the substrate has a large influence on the coating quality. Statistical methods are widely used to quantify the influences of the particle and substrate characteristics. Neural networks provide a greater capability to analyse particle characteristics and substrate temperature data for coating quality control. In this work, the analysis of comprehensive process data and coating characteristics using neural networks is investigated and compared to established design of experiments (DOE) statistical methods.