Topographic indices like the Topographic Wetness Index (TWI) have been used to predict spatial patterns of average groundwater levels and to model the dynamics of the saturated zone during events (e.g., TOPMODEL). However, the assumptions underlying the use of the TWI in hydrological models, of which the most important is that groundwater level variation can be approximated by a series of steady state situations, are rarely tested. It is also not clear how well findings from existing hillslope studies on sites with transmissive soil can be transferred to entire catchments with less permeable soils. This study, therefore, evaluated the suitability of selected topographic indices to describe spatial groundwater level variations based on time series from 51 groundwater wells in a 20 ha catchment with low-permeability soils in Switzerland. Results showed that median groundwater levels were correlated to slope, curvature, and TWI, but the strength of correlation depended on whether the indices characterized the local topography or the topography of the upslope contributing area. The correlation between TWI and groundwater levels was not constant over time but decreased at the beginning of rainfall events, indicating large spatial differences in groundwater responses, and increased after peak flow, when groundwater levels could be considered to be spatially in a steady state. Our findings indicate that topographic indices are useful to predict median groundwater levels in catchments with low-permeability soils and that the TWI assumptions are best met when groundwater levels change slowly.