Newborn screening for classical galactosemia in the Netherlands is performed by five laboratories and is based on the measurement of galactose 1-phosphate-uridyltransferase (GALT) activity and total galactose (TGAL) in heel prick blood spots. Unexpected problems with the GALT assay posed a challenge to switch to a new assay. The aim of this study was to make an analytical and clinical evaluation of GALT assays to replace the current assay and to establish new cut-off values (COVs).First, the manual assay from PerkinElmer (NG-1100) and the GSP assay were compared by analyzing 626 anonymous heel prick samples in parallel. Secondly, a manual GSP method was evaluated and 2,052 samples were compared with the automated GSP assay. Finally, a clinical evaluation was performed by collecting data from 93 referred newborns.No satisfactory correlation was observed between GALT activity measured with the manual NG-1100 assay and the automated GSP assay. An acceptable correlation was found between the manual and automated GSP assay. Intra- and inter-assay variation of the automated GSP were 1.8-10.0% and 3.1-13.9%, respectively. Evaluation of clinical data demonstrated that adjusting the COVs for GALT to 2.0 U/dl and TGAL to 1,100 μmol/l improved specificity of screening for classical galactosemia.An assay designed for automated processing to measure GALT activity in heel prick samples works equally well when processed manually. We therefore adopted both methods in the Dutch screening laboratories. As a result of this evaluation new COVs for GALT and TGAL have been introduced and are valid from July 2015.