Information on the steam and feedwater flow rates in the secondary loop of nuclear power plants is valuable for thermal efficiency estimation and the related controllers in nuclear power plants. However, the high level of noise in measuring flow rates detracts from the usefulness of this information and forces the operator to exclude the values of the steam and feedwater flow rates when controlling the water level of a steam generator at low operating powers. In recent years, it has been proposed that the wavelet transform can reconstruct a signal that approximates very closely the original signal under a high level of noise. A possible way of differentiating the flow rate from noise is proposed by use of the wavelet noise-reduction or denoising technique and, as one of the potential applications for nuclear power plants, the wavelet transform is incorporated into the water-level controller of steam generators for successful control at low operating powers.