Why do some companies employ performance pay while others do not? For performance pay to lead to the desired results, it needs to be based on a suitable performance measure. Therefore, only firms which can accurately measure performance will introduce such a payment scheme. The availability of a performance measure, in turn, largely depends on the type of work conducted in the firm. Understanding work as a bundle of tasks allows one to analyze the content of this work. I argue that the intensity in routine and non-routine tasks can serve as a predictor for performance pay. Regression analysis using survey data delivers mixed evidence for this relationship. The intensity in non-routine tasks which are manual-physical appears to be negatively associated with the probability to employ performance pay, as hypothesized. The intensity in other types of routine or non-routine tasks does not show a significant effect.