The task approach is attracting increasing attention and recognition among scholars in economics, sociology and related fields. However, measurement still presents an important challenge to the task approach. This paper studies the comparability of task measures in the commonly used German BIBB/IAB-BIBB/BAuA employment cross-sections on qualification and working conditions from 1979, 1985/86, 1991/92, 1998/99, and 2006. We hypothesize that findings on task-biased technological change are sensitive to variable choice. The task data differ substantially from other task data, and task items are not readily comparable between survey years. As a result, classifying single task items into distinct domains leads to a number of problems. To test our hypothesis, we use different strategies for classifying tasks into task domains and analyze whether different operationalizations lead to different conclusions about task change in Germany. Our results show that this indeed is the case. Our paper provides readers with a broader understanding of German task data and gives recommendations for applying the task approach to German employment data.