The Open Science Collaboration replicated 100 influential psychology studies and found only 36% held up, with effect sizes half as large. A larger pooled analysis showed 64% eventually replicated but with effects 32% smaller. A 186-researcher collaboration found 50% of 28 classic findings failed even with large samples. The crisis stems from publication bias (97% of positive industry results published vs 39% negative), researcher bias, and fraud cases. Psychology, medicine, and social sciences are all affected.