A chemometric-assisted UV absorption spectroscopic method is proposed for the simultaneous determination of acesulfame-K, aspartame and stevioside in raw powder mixtures of commercial sweeteners. The synergy interval partial least squares (siPLS) algorithm was applied to select the optimum spectral range and their combinations. The utilization of spectral region selection aims to construct better partial least squares (PLS) model than that established from the full-spectrum range. The results show that the siPLS algorithm can find out an optimized combination of spectral regions, yielding lower relative standard error of prediction (RSEP) and root mean square error of prediction (RMSEP), as well as simplifying the model. The RMSEP and RSEP obtained after selection of intervals by siPLS were 0.1330 μg·ml–1 and 1.50 % for acesulfame-K, 0.2540 μg·ml–1 and 1.64 % for aspartame, 1.4041 μg·ml–1 and 2.03 % for stevioside respectively. The recovery values range from 98.12 % to 101.88 % for acesulfame-K, 98.63 % to 102.96% for aspartame, and 96.38 % to 104.04 % for stevioside respectively.