A new method was established for the simultaneous analysis of four homologous benzalkonium chlorides (dodecyldimethylbenzyl ammonium chloride, tetradecyldimethylbenzyl ammonium chloride, hexadecyldimethylbenzyl ammonium chloride, and octadecyldimethylbenzyl ammonium chloride) in compound chemical disinfectants using non-aqueous capillary electrophoresis (CE) based on a micellar electrokinetic chromatography mode with direct ultraviolet detection. The separation was performed on an uncoated fused quartz capillary with a total length of 60.2 cm and a diameter of 25 μm. The separation buffer consisted of a mixture of methanol/acetonitrile (60:40, v/v) containing 70 mmol/L sodium acetate, 60 mmol/L trifluoroacetic acid and 20 mmol/L sodium dodecyl sulfate. The sample buffer was a methanol solution containing only 2 mmol/L trifluoroacetic acid. The separation voltage was set at 8 kV with a working current of approximately 2.3 μA. The detection wavelength was 214 nm. Under optimal conditions, the limit of detection and limit of quantification for these four benzalkonium chlorides (BACs) were 1.0 mg/L and 5.0 mg/L, respectively. Good linearities were observed in the concentration ranges from 5.0 to 100.0 mg/L, with correlation coefficients above 0.999 for all compounds. The recoveries of these four BACs ranged from 92.5 % to 109.1 % with relative standard deviations below 4.7 %. With the new method, all four BACs could be analyzed in a single injection. In contrast, the aqueous CE method in the National Standard GB/T 26369-2020 only allowed for the simultaneous analysis of the first three homologous. The new method demonstrated the improved peak shape compared to the aqueous CE method and then was successfully applied to the analysis of 19 commercially available samples, such as object table disinfectants, hand sanitizers, and disinfectant wipes, which claimed to contain quaternary ammonium compound. The results obtained using the new method were compared with those of the aqueous CE of the National Standard Method, and no statistically significant differences were observed. The new method is simple in pre-treatment and provides accurate results, making it highly suitable for routine analysis.