Spatial distributions of low-permeability soils, such as clay, silt, and mud, are important for establishing hydrogeological maps and can affect groundwater flow and recharge and contaminant transport in soil solutes. This study adopted indicator kriging (IK) to spatially characterize the optimal estimates and uncertainty of the low-permeability soil fraction in the Choushui River alluvial fan aquifers in Taiwan. First, IK was used to analyze the occurrence probabilities of low-permeability soil fractions according to several thresholds and to establish the conditional cumulative distribution function (CCDF) using a linear interpolation model. Then, median estimates and E-type estimates of the low-permeability soil fractions in aquifers were determined based on the CCDF. Finally, the integration of the conditional variance and interquartile range was employed to assess the local uncertainty of IK estimates. The analysis results indicated that the median estimates were more reliable than the E-type estimates and capable of modeling the absence of low-permeability soils in aquifers. Moreover, high estimated low-permeability soil fractions frequently gave rise to the high levels of local uncertainty. The study results are useful for modifying hydrogeological system maps and establishing numerical simulations of groundwater models.