A representative saline-sodic area measuring 300 m×160 m was investigated to determine the salinity and sodicity levels in Lesheng Town, Da'an City, Jilin Province, northeast China on April 20, 2013. Apparent soil electrical conductivity was measured using EM38 (electromagnetic induction) and GPS, and its spatial variability was studied by using classical statistics and geostatistics. Correlations between apparent soil electrical conductivity, including apparent soil electrical conductivity in horizontal direction(EC_h) and apparent soil electrical conductivity in vertical direction (EC_v), and salinity-sodicity parameters, including soil ions, EC_1:5, pH_1:5, sodium adsorption ratio (SAR), and soil salinity content (SC) were analyzed in the saline-sodic upland soil of the Songnen Plain. The results indicated a moderate spatial variability for EC_h and EC_v in classical statistics. Frequency distributions and the Kolmogorv-Smirov test for normality showed that EC_h and EC_v were not normally distributed (P < 0.05). Therefore, mathematical transformations were performed to convert the data to fit the normal distribution, which is a prerequisite for calibration of the theoretical model and generation of semivariogram parameters and kriged maps. After log-transformation, EC_h and EC_v showed normal distributions. The results of geostatistical analyses indicated that EC_h has a strong spatial variability and dependence, and the spatial distribution of EC_h is affected by structural factors, which might include topography, hydrology, and climatic factors. However, EC_v had a moderate spatial variability and dependence, and the spatial distribution of EC_v was jointly affected by structural and random factors. Empirical semivariograms for EC_h were simulated by spherical models, but those of EC_v were simulated by exponential models. The results of a Pearson correlation showed five indexes of salinity-sodicity parameters (pH_1:5, EC_1:5, SAR, SC, Na⁺) were significantly correlated with EC_h and EC_v (P < 0.05). The correlation coefficients between EC_h and five salinity-sodicity parameters were higher than that between EC_v and five salinity-sodicity parameters. A stepwise regression analysis revealed that the regression prediction models with EC_h and EC_v could explain most of the variations of the soil salinity-sodicity parameters. The regression prediction models between apparent soil electrical conductivity and five salinity-sodicity parameters were linear. The determination coefficient of EC_h was higher than that of EC_v. Therefore, the apparent soil electrical conductivity in a horizontal direction could be used to calculate the index of the parameters of soil salinity-sodicity and to indicate soil salinity-sodicity in practical applications.