The NASA-TLX subjective workload assessment is typically used immediately following a participant's performance in an experimental task to assess the workload experienced. However, it is sometimes necessary to assess the anticipated workload of a task before the task is actually performed. This study compares the workload assessments of participants who performed two minimally invasive surgical training tasks to participants who only saw descriptions of the two tasks. Results showed that participants who were asked to rate the anticipated workload underestimated the overall workload required for the "cannulation" task, while overestimating the overall workload required for the rope task. Interactions between task type and condition also were found in three of the NASA-TLX subscales (mental demand, effort, and frustration). Overall, the reliability of participants' prediction of the difficulty depended on the task being evaluated and the particular measure of difficulty that was being assessed. In general, physical facets of workload appeared to be more accurately assessed than cognitive facets.