The *Nature* survey of 1,576 researchers reveals a significant crisis in reproducibility, with over 70% failing to reproduce another scientist's experiments and more than half failing to reproduce their own. Despite this, less than 31% believe that failure to reproduce published results indicates the result is probably wrong, and most still trust the published literature. The survey also highlights contrasting attitudes, with physicists and chemists being the most confident in the reproducibility of their field. While many researchers have tried to publish replication attempts, only a small proportion have been successful. Concrete steps to improve reproducibility, such as redoing experiments and standardizing methods, are being taken by a third of respondents, but these efforts can be time-consuming and costly. The survey suggests that journals, funders, and research institutions should take more action to address reproducibility issues, with nearly 90% of respondents endorsing robust experimental design, better statistics, and improved mentorship.The *Nature* survey of 1,576 researchers reveals a significant crisis in reproducibility, with over 70% failing to reproduce another scientist's experiments and more than half failing to reproduce their own. Despite this, less than 31% believe that failure to reproduce published results indicates the result is probably wrong, and most still trust the published literature. The survey also highlights contrasting attitudes, with physicists and chemists being the most confident in the reproducibility of their field. While many researchers have tried to publish replication attempts, only a small proportion have been successful. Concrete steps to improve reproducibility, such as redoing experiments and standardizing methods, are being taken by a third of respondents, but these efforts can be time-consuming and costly. The survey suggests that journals, funders, and research institutions should take more action to address reproducibility issues, with nearly 90% of respondents endorsing robust experimental design, better statistics, and improved mentorship.