Aquifers in the alluvial or fluvial plains of Taiwan consist of geological materials of different hydraulic conductivity (K). Aquifer heterogeneity concerns the spatial variability of K(x,y,z) due to the spatial distribution of geological materials. Heterogeneity significantly influences groundwater flow fields, distributions of dissolved contaminants, and pathways of nonaqueous phase liquids. Unsuccessful remediation is usually not caused by failure of the physiochemical reactions of the remedial technology employed, but rather due to a lack of understanding of aquifer heterogeneity. Recent scientific research and practical experience all indicate that the low-K zones in a contaminant site are the bottleneck of remediation. Therefore, three-dimensional site investigation techniques should be used in order to effectively delineate locations of the high-K and low-K zones, which require different remediation measures. The multilevel slug test (MLST) is a single-well double packer technique, which is useful in estimating the vertical variation of hydraulic conductivity, K(z), of the geological materials surrounding the test well. Conducting MLSTs at different test wells in a site allows us to determine the three-dimensional K(x,y,z). MLST is particularly suitable for contamination sites as it (1) withdraws no groundwater so requires no consequential contaminated water treatment, (2) produces small pressure disturbance that would not alter the groundwater flow field and contaminant distribution, and (3) is easier and inexpensive to implement. In spite of the popularity MLST, our recent studies indicate that the test section length (the water intake length between the double packers) has significant impact on the estimation of K. In general, a longer test section length reduces the resolution of K(z) obtained, and too long a test screen length introduces too much vertical average effect to the estimates of K, resulting in insignificant depth-specific representation and inaccurate estimates of K(z). Reducing the test section length increases the resolution of K(z) obtained as well as the work load of MLST, and hence the cost of site investigation. The purposes of this project are: to study the influence of the test screen length on the estimates of K, to evaluate the correlation between K(z) resulted from different test section lengths, and to investigate whether or not a standard test section length is possible to yield reliable estimation of K(z) under different hydrogeologic conditions. Results obtained will be useful for an effective performance of MLST, for reducing the cost of site investigation while maintaining the quality of the work, and for the future technical guidance of MLST.