We use rich microdata on bank robberies to estimate individual-level disutilities of imprisonment. The identification rests on the money versus apprehension trade-off that robbers face inside the bank when deciding whether to leave or collect money for an additional minute. The distribution of...