Abstract: Determining internal bond strength and thickness swelling after cyclic aging tests in humid conditions is essential in assessing the moisture resistance of particle and fiber boards. However, as operating procedures for these types of tests take at least three weeks, their use in daily finished product control is impracticable. To solve this problem an artificial neural network was used as a predictive method to determine these values from the board properties of thickness, density and moisture content in conjunction with thickness swelling and internal bond strength values obtained before the aging cycle. Using 113 boards, an artificial neural network was designed consisting of two separate feedforward multilayer perceptrons, applying the hyperbolic tangent as the transfer function. Training was conducted through supervised learning after the input data had been normalized. In the testing group the network attained a determination coefficient of 0.94 for the internal bond strength test and 0.92 for the thickness swelling test.