Tests of local realism based on Bell’s inequality have, up to now, always had a small loophole, but that has now been closed. Lynden K Shalm of NIST in Boulder, Colorado, US, and colleagues looked at entangled photon pairs, ensuring that the events used were space-like separated and using fast random-number generators and high-speed polarisation measurements. The results: local realism is rejected with a p value of 5.9 × 10–9.