After brain injury, people often suffer from temporary or permanent motor problems. This makes it more difficult to perform everyday activities, such as making a sandwich. Imagining movements (called motor imagery) or learning by observing movements could potentially contribute to motor recovery. The hypothesis is that people partly use the same brain networks during execution, observation and imagery of movements. However, little is known about whether people are still able to imagine movements after a stroke, and how and whether motor imagery recovers over time. For example, it is unclear what percentage of patients after a stroke show reduced performance on motor imagery tasks and how different imagery tasks correlate with each other. Also, little is known about the longitudinal course of motor imagery performance after a stroke. This dissertation focuses on whether and how mental imagery, and motor imagery in particular, is affected after a stroke. In this thesis, the effects of stroke on a number of imagery tests and tasks were investigated in a rehabilitation setting. The results show that a large proportion of patients can still (implicitly) imagine movements, that there are individual differences between patients on these tests, and that motor imagery ability can recover over time. These results argue for the use of multiple types of imagery instruments for the screening, selection and monitoring of stroke patients.