This paper uses the National Industrial Recovery Act of 1933, which set up industry-wide cartels in the manufacturing sector of the US economy, to gain empirical insight into the current debate on the output effects of cartels. Recent theoretical studies have demonstrated ways in which cartels could expand, rather than reduce output as is traditionally thought. The New Deal cartel experiment does not support this ‘efficient cartel’ view. On the contrary, the legislation brought about a reduction in manufacturing output, as traditional cartel theory would predict.