This paper uses the cartel-enabling National Industrial Recovery Act (NIRA) of 1933 to gain insight into cartel performance. I employ a monthly panel of 66 industries that passed an NIRA code of fair competition to examine how specific attributes of these cartel codes affected the ability to achieve collusive outcomes. I find that output growth was significantly lower during cartel months, consistent with cartel theory, and that industries with more complex codes were more successful than those with simpler ones. Furthermore, industries with code restrictions on new productive capacity, production quotas, and requirements to file data with a central board were die most successful at reducing output, which suggests that these types of provisions were the most effective in helping firms attain collusive outcomes. Finally, I find that the effectiveness of data-filing provisions was limited to the early months of the NIRA, prior to a wave of cartel breakdown occurring in spring 1934.