We present a rapid, robust method of signal processing useful for optoacoustic monitoring of total hemoglobin concentration ([THb]) and oxygen saturation level in small blood vessels. Our method includes the wavelet-based regularization of the difference operator which is a typical discrete approximation of the derivative. The optimal degree of regularization is defined by the signal-to-noise ratio (SNR). We applied the proposed method to Monte Carlo-modeled signals from a cylinder simulating the human radial artery (diameter 1.6 mm, depth from skin 2 mm, and [THb] varied in a wide range from 4-16 g/dL). We obtained N-shaped signals and found that the maximum of the first derivative between the front and rear walls systematically correlates with the actual value of [THb]. We estimated the accuracy of [THb] reconstruction from the maximum of the first derivative as 0.32 ± 0.18 g/dL (mean value ±SD) at an SNR typical for our in vivo experiments at the wavelength of 1064 nm. We also demonstrated that the difference between the maxima of the first derivative of the signals obtained at 700 nm and 1000 nm depends on oxygen saturation level.