Current climate change models for the southeast UK predict changing rainfall patterns, with increased incidence of extreme events. The chalk aquifer in the UK and northern France is susceptible to groundwater-induced flooding under such conditions. In this methodological study we apply a frequency domain analysis approach to the chalk aquifer to derive a transfer function between effective rainfall and groundwater level from 7 years of monitoring data from the North Heath Barn site, near Brighton. The derived transfer function was calibrated and validated against monitoring data and then used to predict groundwater level for rainfall models for high, medium and low emission scenarios from the UKCP09 database. The derived transfer function is most closely comparable to the linear aquifer model, despite evidence for both matrix and fracture or karst water flow in the chalk, with transmissivity and unconfined storativity at the catchment scale of 1548 m2 day−1 and 1.6 × 10−2. The application of the transfer function to UKCP09 rainfall data suggests that groundwater-induced flooding may be about four times more frequent by 2040–2069 compared with 1961–1990 and seven times more frequent by 2070–2099. The model data also suggest an increase in the duration of groundwater minima relative to the reference period. Compared to deterministic modelling which requires detailed knowledge of aquifer heterogeneity and processes, the transfer function approach, although with limitations, is simpler, incorporating these factors into the analysis through frequency and phase coefficients, and thus may have the potential for groundwater risk assessment in other areas. Copyright © 2015 John Wiley & Sons, Ltd.