The safety of a geologic repository for heat-generating nuclear waste relies in part on the long-term performance of the clay-based buffer material that surrounds the waste canisters. The sealing and waste-isolation properties of the bentonite may degrade as temperatures rise to high values. To improve the understanding of the evolution of the repository's near field and in particular the effects of increased temperatures on the bentonite buffer, a full-scale, long-duration experiment named HotBENT is underway at the Grimsel Test Site, Switzerland. The HotBENT monitoring system was designed with the support of a modelling-based data-worth analysis in which the contribution of individual sensors to the reduction in estimation or prediction uncertainties is formally evaluated. Two examples are described; the first demonstrates optimization of the monitoring system for the estimation of parameters of a clay-swelling model, whereas the second example shows how to select monitoring data that help improve the quality of model predictions needed for decision support. A data-worth analysis was considered a useful approach to identify the information content of actual and planned monitoring data for the specific objectives of the experiment. The HotBENT monitoring system was eventually designed in part based on the results presented in this paper. Thematic collection: This article is part of the Sustainable geological disposal and containment of radioactive waste collection available at: https://www.lyellcollection.org/topic/collections/radioactive