This work investigates the proportion of generated fines in a pilot-scale experiment using a belt conveyor and commercial fuel pellets. For this, a belt conveyor with a length of 3.1 m was used and operated at varying conditions: speeds, percentages of material loading on the belt, two combinations of the inclination angle of the belt and the falling height, and a different number of handling steps. We considered a design of experiments approach based on response surface methodology to investigate the effect of different conditions on the potential of fines generation. Moreover, a comparison between the results of the belt conveyor and three common benchmark experimental approaches (tumbling box, rotary impact tester, and mechanical compression test) was made. Results show that the number of handling steps and the combined effect of drop height and inclination angle directly affected the fines generation. However, the tested belt speed range and the level of loading were of lower significance. A polynomial quadratic model was derived based on the regression analysis and showed a high accuracy to predict the proportion of fines. Moreover, the tumbling box method showed good potential to predict the proportion of fines in a belt conveyor when transported several times.