Predictions of macropore flow is important for maintaining both soil and water quality as it governs key related soil processes e.g. soil erosion and subsurface transport of pollutants. However, macropore flow currently cannot be reliably predicted at the field scale because of inherently large spatial variability. The aim of this study was to perform field scale characterization of macropore flow and investigate the predictive performance of (1) current empirical models for both water and air flow, and (2) X-ray CT derived macropore network characteristics. For this purpose, 65 cylindrical soil columns (6 cm diameter and 3.5 cm height) were extracted from the topsoil (5 to 8.5 cm depth) in a 15 m × 15 m grid from an agricultural loamy field located in Silstrup, Denmark. All soil columns were scanned with an industrial CT scanner (129 mum resolution) and later used for measurements of saturated water permeability, air permeability and gas diffusivity at -30 and -100 cm matric potentials. Distribution maps for both water and air permeabilities and gas diffusivity reflected no spatial correlation irrespective of the soil texture and organic matter maps. Empirical predictive models for both water and air permeabilities showed poor performance as they were not able to realistically capture macropore flow because of poor correlations with soil texture and bulk density. The tested empirical model predicted well gas diffusivity at -100 cm matric potential, but relatively failed at -30 cm matric potential particularly for samples with biopore flow. Image segmentation output of the four employed methods was nearly the same, and matched well with measured air-filled porosity at -30 cm matric potential. Many of the CT derived macropore network characteristics were strongly interrelated. Most of the macropore network characteristics were also strongly correlated with saturated water permeability, air permeability, and gas diffusivity. The correlations between macropore network characteristics and macropore flow parameters were further improved on dividing soil samples into samples with biopore and matrix flow. Observed strong correlations between macropore network characteristics and macropore flow highlighted the need of further research on numerical simulations of macropore flow based on X-ray CT images. This could pave the way for the digital soil physics laboratory in the future.