Field research was conducted to determine the spatial variability of saturated hydraulic conductivity on four central Pennsylvania  soils. The soils were classified as: a coarse-loamy, siliceous, mesic, Entic Haplorthod (Gatesburg loamy sand); a fine, mixed, mesic, Typic Hapludalf (Hagerstown silt loam); a fine-loamy, mixed, mesic, Typic Fragiudult (Monogahela silt loam); and a finesilty, mixed, mesic, Dystric Fluventic Eutrochrept (Nolin silt loam).  Saturated hydraulic conductivity was measured in-situ at 1-m intervals on a,40-m transect for each soil at a depth of 0.5-0.6 m using the Guelph Permeameter method. Considerable spatial  variation in measured values of field saturated hydraulic conductivity was found  along the transect in all four soils. A stationarity test was employed on raw data and a significant trend was found along the transect for three of the four soils. The trend was removed using a third-order polynomial on the data of the Hagerstown soil and on the log-transformed data of the Monongahela soil.  A sine function fit the data of the Nolin soil. Geostatistical methods were used to assess the spatial structure among the measurements. The semivariograms and autocorrelograms revealed a lack of spatial structure in the detrended data (residuals). A significant spatial structure described by an exponential model existed up to about 5 m in the original data of the Gatesburg soil which did not have a trend, and to at least 40 m in the other three soils as a result of the trend. These results indicate that many soil mapping units will contain trends in hydraulic conductivity which must be evaluated to characterize an area adequately.