An approach for deriving pulse rate variability (PRV) from photoplethysmography (PPG) signals is proposed. By combining sliding window iterative discrete Fourier transform (DFT) with the Hilbert transform algorithm, this method effectively reduces the influence of noise and sampling frequency compared with those of traditional methods. First, the fundamental component of the PPG signal is computed with the sliding window iterative DFT algorithm. Then, it is processed by an integer coefficient low-pass filter and the Hilbert transform to produce an instantaneous pulse rate (IPR). Finally, PRV is extracted from IPR in the frequency domain. PRV is also derived from the PPG signal in the time domain for comparison with that derived using the proposed method. Furthermore, PRV obtained from PPG signals at various sampling frequencies and with various noise levels is investigated. The results show that the proposed method can accurately derive PRV from PPG signals, even if the PPG signal has a low sampling frequency (4 Hz) and high noise (e.g., SNR is about 3.0). The experimental results indicate that the proposed method can be used to estimate PRV when the subjects are in different states (rest, sleeping, or visual fatigue). Moreover, the proposed method is efficient and thus suitable for detecting PRV in real time. It has potential for PRV assessment in various medical fields, such as home health and clinical and hospital environments.