In telemanipulation commonly master and slave have dissimilar workspaces. Workspace extension methods can overcome this mismatch between master and slave. Literature proposes several workspace extension methods for translations such as scaling and indexing. However, for rotations it is unclear how workspace extension methods should be designed. The present study proposes a methodology to design rotational workspace extension methods with a variable gain. Which is designed based on the speed-accuracy trade-off and several task characteristics, like the distribution of rotational amplitudes during telemanipulation tasks. The effectiveness of the variable gain method is evaluated in terms of task performance and control effort in a within-subject-haptic-telemanipulation-single-degree-rotational-pointing-experiment, based on Fitts’ tapping task. The parameters are chosen in accordance with a care robot case study where the rotational workspace of the master device is 45°, but where most tasks require 90° slave rotation, and some even 180°. It is hypothesized that variable gain workspace extension allows improved performance in regions it is customized for (up to 90% of the slave rotations) with respect to a conventional constant scaling, while the operator is able to perform in all regions with similar order of magnitude metrics. To test this hypothesis a variable scaling method, a constant scaling method, and a baseline method (without scaling) are designed. The experimental results show improved performance on fine positioning time and reversal rate for the variable scaling method at the focus region. Furthermore, human operators accept variable scaling and are able to manipulate linear changes of the gain equally smooth as constant gains, while high nonlinear changes of the gain are more difficult to manipulate smoothly. To conclude, this study demonstrates a methodology for designing variable gain workspace extension methods for specific task characteristics which allows improved execution performance compared to the conventional constant scaling method. For applying this methodology in real-life applications, the results need to be scaled for more realistic situations, such as higher degrees of freedom and in-contact tasks.