Personalized medicine, otherwise called stratified or precision medicine, aims to better target intervention to the individual to maximize benefit and minimize harm. This review discusses how diabetes aetiology, pathophysiology and patient genotype influence response to or side effects of the commonly used diabetes treatments. C-peptide is a useful biomarker that is underused to guide treatment choice, severe insulin deficiency predicts non-response to glucagon-like peptide-1 receptor agonists, and thiazolidinediones are more effective in insulin-resistant patients. The field of pharmacogenetics is now yielding clinically important results, with three examples outlined: sulphonylurea sensitivity in patients with HNF1A maturity-onset diabetes of the young; sulphonylurea sensitivity in patients with Type 2 diabetes with reduced function alleles at CYP2C9, resulting in reduced metabolism of sulphonylureas; and severe metformin intolerance associated with reduced function organic cation transporter 1 (OCT1) variants, exacerbated by drugs that also inhibit OCT1. Genome-wide approaches and the potential of other ‘omics’, including metagenomics and metabolomics, are then outlined, highlighting the complex interacting networks that we need to understand before we can truly personalize diabetes treatments.