A previous literature cautions that paying workers for performance might crowd out non-monetary motives to work hard. Empirical evidence from the field, however, has been based on between-subjects designs that are best suited for detecting crowding out due to low-powered incentives. High-powered incentives in the workplace tend to increase output, but it is unknown whether this masks crowding out. This paper uses a within-subject experimental design and finds evidence that crowding out also extends to high-powered incentives, in a real work setting with paid workers. There is individual heterogeneity, however, with a minority of workers report crowding in of motivation. Thus, the impact of performance pay might depend on the mix of worker types.