An important potential application of the electrical-resistivity method occurs in salinity studies of lakes and water reservoirs. However, uncertainty exists because there is a problem of equivalence in resistivity data interpretation and because the resistivity variation in the water layer in reservoirs is subtle (contrasts of a factor of two) and changes over short intervals of time (typically in hours or days). We carry out numerical modeling and inversion of synthetic data and field resistivity data from Lake Whitney, Texas, U.S.A., to examine how accurately resistivity in the water column in reservoirs can be determined using the electrical-resistivity method. Our objective is to advance the method as a tool in limnologic research for mapping freshwater zones in impacted lakes and water reservoirs. The simulated freshwater target in the synthetic data effectively is realized from the inversion with root-mean-square (rms) error less than 10%. However, the resolutionof the inverted sections decreases with increasing noise. Inversions of the field apparent-resistivity data from three profiles in the lake, computed using estimated optimization parameters from the synthetic study, reveal the possible pattern of salinity distributions in the reservoir. For unconstrained inversion schemes, comparisons of the inverted and independently measured in situ water electrical-conductivity data yield an average rms error of 10.8%. This error value is reduced to approximately 5% with inclusions of a priori information on water resistivity in the inversion scheme. We observe an inverse relationship between error level and number of constraints on water resistivity. In general, results show that the electrical-resistivity method is a viable tool for mapping salinity variations in reservoirs. However, good data quality and inclusion of measured water conductivity as constraints in the inversion schemes are important to enhance accuracy of the inversion results.