Estimating moisture content of the grain on the combine provides basic information for Precision Agriculture, yet the apparatus is more complicated than those used for grain weight monitoring. The method presented in this paper is very simple and to differentiate the change in the impact sound of the grain in the tank associated with its moisture content. The sensor consisted of a small steel plate to receive the impact of the emitted grain and an earphone was adhered behind the plate to generate the signal. Each time the sensor received the impact, the signal was processed with Fast Fourier Transform (FFT) and the spectra were obtained up to 10 kHz. The grain moisture content of paddy rice and wheat kernel was then estimated from linear multiple regression of the amplitudes at two frequencies. The standard errors were 0.36% for paddy rice (ranging 19.0% - 23.1% w.b.) and 0.77% for wheat kernel (ranging 17.5% - 29.5% w.b.).