This paper describes the processing and characterization of epoxy matrix composites reinforced withshort flakes obtained from the scales of a typical fresh water fish (
Lobea rohita
) and also reports on theirerosion wear characteristics. Composites are developed by reinforcement of randomly oriented shortflakes obtained from fish scales into epoxy resin. Erosion characteristics are studied with the help of an air jet type erosion test rig employing the design of experiments approach based on Taguchi’s orthog-onal arrays. The findingsof theexperimentsindicate that therate of erosionbyimpact of solid erodent isgreatly influenced by various control factors. The experimental results are in good agreement with thevalues from the theoretical model. An artificial neural network (ANN) approach is also applied to predictthe wear rate of the composites and compared with the theoretical results. An optimal parameter com-binationis determined, whichleads tominimizationof erosionrate. Analysis of variance(ANOVA) is per-formedonthemeasureddataandS/N(signaltonoise)ratios.Amathematicalcorrelation,consistentwiththeexperimental observations isproposedas apredictiveequationfor estimationof erosionrateof thesecomposites. Finally, popular evolutionary approach known genetic algorithm (GA) is used to generalizethe method of finding out optimal factor settings for minimum wear rate.