Readout signal and noise power spectrum of an optical disk system were simulated and the results were compared with the measured power spectrum to analyze the major sources of noise in the system. The readout signal and power spectrum were calculated by varying the mark shape and position as well as beam off-track range. The variations of mark length, position and beam off-track range were assumed to have a Gaussian probability distribution. The power spectrum of the readout signal was calculated by executing a discrete fast Fourier transform algorithm. From the fitting of the measured signal power spectrum, the major sources of noise in our optical disk system were deduced to be the variations of mark length, position and beam off-track range.