The influence of column selection on the determination results of 23 preservatives in cosmetics was studied by considering the corresponding detection methods for these preservatives. The test method for 23 preservatives specified in Safety and Technical Standards for Cosmetics was considered as the reference method. Twenty three preservatives were analyzed using 15 different brands and models of C18 chromatographic columns on two different high performance liquid chromatographs. The plate count and resolution of the chromatographic peaks were calculated to analyze the separation effect of the 23 components. The separation efficiency and equivalence of different chromatographic columns were evaluated and predicted using the database established by the United States Pharmacopeia (USP) and Product Quality Research Institute (PQRI). The experimental results showed that the separation effects of the 15 chromatographic columns on the 23 preservatives differed significantly. Further, only two chromatographic columns were able to completely separate the 23 components without changing the standard procedure for mobile phase elution and other conditions. The other 13 columns poorly separated some components. The retention times of phenyl 4-hydroxybenzoate, isobutyl 4-hydroxybenzoate, butyl 4-hydroxybenzoate, and benzyl 4-hydroxybenzoate were similar. The separation of these substances was among the key outcomes of the chromatographic separation. The differences in the selectivity of the 4-hydroxybenzoates were distinct among the different chromatographic columns. The plate counts of the 4-hydroxybenzoates were similar when using the same chromatographic column on two high performance liquid chromatographs. However, the plate counts of 4-hydroxybenzoates using different chromatographic columns were considerably different. Control of the plate counts alone was not sufficient to achieve good separation efficiency of the 4-hydroxybenzoates. Moreover, the peak sequences of some compounds such as 4-hydroxybenzoic acid and methylisothiazolinone with different chromatographic columns were reversed. Some substances could be absorbed at multiple wavelengths. If separation from adjacent chromatographic peaks is not achieved at the baseline, large errors will occur when detecting the low-concentration components to be tested. The ZORBAX SB-C18 column was used as a reference column, and the similarity values provided by the USP and the PQRI databases were consulted. The actual separation efficiency of the typical column used in the experiment was compared with the predicted results of selectivity difference from the database. The results showed that the USP and the PQRI database could not predict a suitable equivalent column. There was no regularity in the experimental results; further, the USP and the PQRI datebase provided scarce reference values for the analysis of the 23 preservatives by liquid chromatography. The outcomes obtained using liquid chromatography methods are easily affected by factors such as the instrument model, mobile phase composition and pH, chromatographic column, column temperature, and flow rate. The chromatographic column is a key factor affecting the accuracy of determination by HPLC. The selectivity difference of the chromatographic column should be taken into account when the technique is performed in relevant laboratories. The standard technique will be employed by different laboratories at different times and under different test conditions, and many abnormalities or deviations may occur. The sequence of peaks should be confirmed using a single compound standard during the first standard method validation or during the replacement of columns. Based on existing research results, future research should focus on the development of a screening and prediction evaluation system for chromatographic columns, and thereby guide the separation of actual samples in complex cosmetic bases. Relevant laboratories should focus on the durability of chromatographic columns, and improve the system adaptability index when developing a detection method. They also should refine the column classification and enhance descriptive information, and thus, guide the rational selection of the column so as to mitigate the risk posed by inaccurate determination results due to improper selection of the chromatographic column.