Objective To determine the association between legal recreational cannabis sales (hereafter ‘legalisation’) and cardiovascular hospitalisations. This study leverages a natural experiment wherein legalisation began earlier in the Western census region of the United States (January 2014) versus non-Western regions (November 2018). Design Using the National Inpatient Sample (NIS), we performed a difference-in-differences analysis of the association between legalisation and hospitalisation rates for six cardiovascular diagnoses, a positive control (cannabis-related diagnoses) and two negative control diagnoses (opioid-related and urinary stones). Setting US states before (2012–2013) and after (2014–2018) legalisation in the Western region Participants Hospitalised individuals in the Western region had a median age of 51 years, 57% were female and 43% were non-white. In other regions, the median age was 55 years, 57% were female and 33% were non-white. Exposure Legal recreational cannabis sales. Main outcomes and measurements Annual hospitalisations per capita for atrial fibrillation/flutter (AF), heart failure (HF), myocardial infarction (MI), ischaemic stroke, venous thromboembolism (VTE) and ventricular tachycardia (VT). Results In this fully adjusted analysis, each year of legalisation was associated with the following relative change in hospitalisation rates expressed as change in the incidence rate ratio (∆IRR): cannabis-related diagnoses (∆IRR 1.138, 95% CI 1.066 to 1.214, p<0.001), AF (∆IRR 1.023, 95% CI 1.010 to 1.035, p<0.001), HF (∆IRR 1.037, 95% CI 1.029 to 1.046, p<0.001), MI (∆IRR 1.031, 95% CI 1.014 to 1.048, p<0.001), ischaemic stroke (∆IRR 1.011, 95% CI 0.999 to 1.024, p=0.078), VTE (∆IRR 1.009, 95% CI 0.999 to 1.020, p=0.066) and VT (∆IRR 1.019, 95% CI 0.993 to 1.046, p=0.161). For falsification testing, there was no change in hospitalisations for urinary stones (∆IRR 1.006, 95% CI 0.986 to 1.027, p=0.535) or opioid-related diagnoses (∆IRR 0.966, 95% CI 0.927 to 1.006, p=0.091). Conclusions Legalisation was associated with increased hospitalisations for AF, HF and MI. Global legalisation trends may result in epidemiologically and economically important increases in cardiovascular healthcare use. Data may be obtained from a third party and are not publicly available. All data are from the National Inpatient Sample. These data are available through the Healthcare Cost and Utilisation Project (HCUP) upon registration and completion of a Data Use Agreement. Access details are available at www.hcup-us.ahrq.gov.