The HDL hypothesis has suffered damage in the past few years. Clinical trials have shown that raising HDL cholesterol levels does not improve cardiovascular disease (CVD) outcomes. In addition, Mendelian randomization studies have shown that DNA variants that alter HDL cholesterol levels in populations are unrelated to incident CVD events. Balancing this deluge of negative data are substantial basic science data supporting the concept that raising HDL cholesterol levels reduces CVD risk. Also, functionally relevant HDL subfractions might be more important determinants of risk than overall HDL cholesterol levels. But, while wobbly, the HDL hypothesis is still standing, seemingly too big to fail owing to past intellectual, economic and psychological investments in the idea.