I study how scientists respond to interruptions in the flow of their research funding, focusing on research grants at the National Institutes of Health (NIH), which awards multi-year grants that researchers can renew. However, there can be delays during the renewal process. In a two-year period around these delays, I find that interrupted labs reduce spending on inputs by over 20% in an average month but over 90% at the lowest point. This change in spending is mostly driven by a decrease in payments to employees that is partially mitigated when scientists have other grants to draw on.