Objectives To evaluate the relationship between FAST peak percentage by adapted Bio-Rad Vnbs analysis using the valley-to-valley integration and genotypes with the aim to improve differentiation between severe a-thalassaemia forms (HbH disease) and the milder disease types. Method DNA analysis for a-thalassaemia was performed on 91 dried blood spot samples presenting normal and elevated FAST peak levels, selected during three years of Dutch national newborn screening. Results Significant differences were found between samples with and without a-thalassaemia mutations, regardless of the genetic profiles. No significant difference was demonstrated between HPLC in 2a/aa and 2a/2a, between 2a/2a and --/aa and between --/aa and --/2a genotypes. Conclusion This study confirms that the percentage HbBart's, as depicted by the FAST peak, is only a relative indication for the number of a genes affected in a-thalassaemia. Based on the data obtained using the modified Bio-Rad Vnbs software, we adopted a cut-off value of 22.5% to discriminate between possible severe a-thalassaemia or HbH disease and other a-thalassaemia phenotypes. Retrospectively, if this cut-off value was utilized during this initial three-year period of neonatal screening, the positive predictive value would have been 0.030 instead of 0.014.