Many tests have been established as valuable for the diagnosis or monitoring of disease but have failed on evaluation for population screening to show any improvement in key outcomes (survival or health-related quality of life). Studies in population biobanks indicate much weaker gene penetrance (ie, association with overt disease) than the widely cited estimates derived from clinically ascertained families presenting with disease.3 Disease penetrance is variable because even in the presence of a rare causative pathogenic variant, other (genetic and environmental) factors contribute—often substantially—to whether, when, and how severely a disease manifests.4 Furthermore, different rare pathogenic variants in the same gene can have widely differing phenotypic severity.5 Even amalgamating data worldwide, population biobanks and databases are currently too small or biased to allow accurate estimation of population-level disease penetrance for most genes, let alone variant-specific risks. ...heritability is 31% for female breast cancer, 15% for colorectal cancer, and 38% for male myocardial infarction.6,7 Furthermore, the PRS report on only a modest fraction of this genetic component: even with larger genome-wide association studies, the majority of disease-associated variants will continue to escape discovery on account of being too rare or of too small an effect size. The health service resources required for follow-up and the potential for parental anxiety have not yet been evaluated in such settings, nor have the sequelae for family members who test positive on cascade screening.11 Testing for many diseases in a single assay might be a false economy A major sell of WGS at birth or PRS for common diseases is the economy of scale of delivering multiple tests in a single assay.12 However, as Wilson and Jungner emphasised, before population screening is introduced for any disease, careful scrutiny of the disease's natural history, of whether presymptomatic detection and intervention genuinely improves survival, and of the performance characteristics in the general population of the proposed test, is required.2 The danger of a technology-centric approach (rather than a disease-centric one) is that it may encourage inclusion in population screening of diseases for which there is no benefit from screening, as well as use of genomic approaches for diseases for which superior screening tests exist.