Characterization of varying bedrock properties is a common need in various contexts, ranging from large infrastructure pre-investigations to environmental protection. A direct current resistivity and time domain induced polarization (IP) survey aiming to characterize properties of a Cretaceous limestone was carried out in the Kristianstad basin, Sweden. The time domain IP data was processed with a recently developed method in order to suppress noise from the challenging urban setting in the survey area. The processing also enabled extraction of early decay times resulting in broader spectra of the time decays and inversion for Cole–Cole parameters. The aims of this study is to investigate if large-scale geoelectrical variations as well as small-scale structural and compositional variations exist within the Kristianstad limestone, and to evaluate the usefulness of Cole–Cole inverted IP data in early time ranges for bedrock characterization. The inverted sections showed variations within the limestone that could be caused by variations in texture and composition. Samples from a deep drilling in the Kristianstad basin were investigated with scanning electron microscopy and energy dispersive X-ray spectroscopy, and the results showed that varying amounts of pyrite, glauconite and clay matrix were present at different levels in the limestone. The local high IP anomalies in the limestone could be caused by these minerals otherwise the IP responses were generally weak. There were also differences in the texture of the limestone at different levels, governed by fossil shapes and composition, proportions of calcareous cement and matrix as well as amount of silicate grains. Textural variations may have implications on the variation in Cole–Cole relaxation time and frequency factor. However, more research is needed in order to directly connect microgeometrical properties in limestone to spectral IP responses. The results from this study show that it is possible to recover useable spectral information from early decay times. We also show that under certain conditions (e.g. relatively short relaxation times in the subsurface), it is possible to extract spectral information from time domain IP data measured with on-off times as short as 1 s.