Background: The demand for vitamin D testing is increasing in China. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) immunoassay is specific and accurate but requires expensive equipment, experienced operators, and complicated pretreatment of serum. Automated immunoassays are simple and convenient but only determine total 25-hydroxyvitamin D (25OHD). The objective of this study was to quantify 25OHD(2) and 25OHD(3) in patients to assist clinical physicians and laboratory directors in choosing the most appropriate method to determine 25OHD. Methods: Vitamin D testing was conducted for 23,695 patients in Peking Union Medical College Hospital from May 2015 to January 2017. Using this large data set, the prevalence and levels of 25OHD(2) were analyzed. LC-MS/MS was used to separately determine 25OHD(2) and 25OHD(3). Results: 25OHD(2) (>= 2.5 ng/mL) was detected in 16.4% (3877/23,695) of patients. Males had a significantly lower incidence of detectable 25OHD(2) (p < 0.01); 1077 (13.9%) samples contained detectable 25OHD(2) (median: 3.7 ng/mL; 2.5%-97.5%: 2.5-17.2 ng/mL). For females, 2800 (17.5%) samples contained detectable 25OHD(2) (median: 4.0 ng/mL; range: 2.5-20.6 ng/mL). Of the 3877 patients with detectable 25OHD(2), males had a significantly higher level of 25OHD(3) (p < 0.01). There was no significant difference in total 25OHD. The proportion of 25OHD(2) in total 25OHD was 1.3%-100%; 87.5% (3391/3877) of the samples contained <10 ng/mL 25OHD(2). 25OHD(2) negatively correlated with 25OHD(3) (r = -0.197, p < 0.01) and positively correlated with total 25OHD (r = 0.217, p < 0.01). Conclusions: Prevalence of 25OHD(2) in patients tested for vitamin D is relatively high in China. 25OHD(2) is significantly negatively correlated with 25OHD(3).