Objectives Multiple 25(OH)D assays have recently been aligned to improve comparibility. In this study we investigated the performance of these assays using both native single-donor sera with target values certified by a reference method as well as single donor sera from a heterogeneous patient population. Design & methods 25(OH)D levels were measured in twenty reference samples (Ref!25OHD; Labquality, Finland) using five automated methods (Lumipulse, Liaison, Cobas, iSYS and Access) and one aligned ID-XLC-MS/MS method (slope: 1,00; intercept: 0,00; R = 0,996). Furthermore, 25(OH)D concentrations measured in 50 pregnant women and 52 random patients using the 5 automated assays were compared to the ID-XLC-MS/MS. In addition, Vitamin D binding protein (DBP) was measured. Results Most automated assays showed significant differences in 25(OH)D levels measured in reference samples. Slopes varied from 1,00 to 1,33, intercepts from − 5.48 to − 15,81 nmol/L and the R from 0,971 to 0,997. This inaccuracy was even more prominent in a heterogeneous patient population. Slopes varied from 0,75 to 1,35, intercepts from − 9.02 to 11,51 nmol/L and the R from 0,840 to 0,949. For most assays the deviation in 25(OH)D concentration increased with elevating DBP concentrations suggesting that DBP might be one of the factors contributing to the inaccuracy in currently used automated 25(OH)D methods. Conclusions Despite the use of standardized assays, we observed significant differences in 25(OH)D concentrations in some automated methods using reference material obtained from healthy single donor sera. In sera of a patient population this inaccuracy was even worse which is highly concerning as patient samples are being investigated in clinical laboratories.