The hallmark of good science is often supposed to be experiments that produce the same results when repeated. But over the last number of years; scientists have replicated a number of established; high-profile experiments and produced different results. Does it point to serious flaws and biases in the sciences? Or is it evidence of the power of science to self-correct? And what can be done to make science more replicable? We explore whether the replication crisis undermines our trust in science.