In particle size measurement using dynamic light scattering (DLS), noise makes the estimation of the particle size distribution (PSD) from the autocorrelation function data unreliable, and a regularization technique is usually required to estimate a reasonable PSD. In this paper, we propose an Lp-norm-residual constrained regularization model for the estimation of the PSD from DLS data based on the Lp norm of the fitting residual. Our model is a generalization of the existing, commonly used L2-norm-residual-based regularization methods such as CONTIN and constrained Tikhonov regularization. The estimation of PSDs by the proposed model, using different Lp norms of the fitting residual for p = 1, 2, 10, and ∞, is studied and their performance is determined using simulated and experimental data. Results show that our proposed model with p = 1 is less sensitive to noise and improves stability and accuracy in the estimation of PSDs for unimodal and bimodal systems. The model with p = 1 is particularly applicable to the noisy or bimodal PSD cases.