Introduction Thrombolysis usage in ischaemic stroke varies across sites. Divergent advice from professional guidelines and product labels may contribute. Patients and methods We analysed SITS-International registry patients enrolled January 2010 through June 2016. We grouped sites into organisational tertiles by number of patients arriving 2.5h and treated 3h, percentage arriving 2.5h and treated 3h, and numbers treated 3h. We assigned scores of 1-3 (lower/middle/upper) per variable and 2 for onsite thrombectomy. We classified sites as lower efficiency (summed scores 3-5), medium efficiency (6-8) or higher efficiency (9-11). Sites were also grouped by adherence with European product label and ESO guideline: label adherent' (>95% on-label), guideline adherent' (5% off-label, 95% on-guideline) or guideline non-adherent' (>5% off-guideline). We cross-tabulated site-efficiency and adherence. We estimated the potential benefit of universally selecting by ESO guidance, using onset-to-treatment time-specific numbers needed to treat for day 90 mRS 0-1. Results A total of 56,689 patients at 597 sites were included: 163 sites were higher efficiency, 204 medium efficiency and 230 lower efficiency. Fifty-six sites were label adherent', 204 guideline adherent' and 337 guideline non-adherent'. There were strong associations between site-efficiency and adherence (P<0.001). Almost all label adherent' sites (55, 98%) were lower efficiency. If all patients were treated by ESO guidelines, an additional 17,031 would receive alteplase, which translates into 1922 more patients with favourable three-month outcomes. Discussion Adherence with product labels is highest in lower efficiency sites. Closer alignment with professional guidelines would increase patients treated and favourable outcomes. Conclusion Product labels should be revised to allow treatment of patients 4.5h from onset and aged 80 years.