Electronic health records (EHRs) coupled to DNA biobanks are potentially powerful but untested resources for pharmacogenomic discoveries. As described in this issue, Delaney and co-workers validated the use of EHRs by demonstrating that loss-of-function CYP2C19*2 was associated with poorer cardiovascular outcomes in clopidogrel-treated patients with an effect size similar to that reported in more controlled clinical trials. Whether studies from real-world EHR-biobanks will supplant randomized clinical trials to provide a sufficient evidence base to change practice is less certain.