© 2018 Taylor & Francis. The aim of this study was to generate models to predict the salinity of the soils in the geothermal zone which is located in Michoacán, México, over space and time. Systematic sampling was performed at locations 400 m apart with a sampling period every 48 days. At each site, two soil samples, from depths of 0–30 cm and 30–60 cm, were collected and analysed in laboratory. Regression-Kriging was used to make spatial predictions of the electrical conductivity of soil (ECs). Spherical and Gaussian theoretical models fit the experimental semivariograms. The results showed that the soil samples ranged from slightly saline (2.1–4 dS m−1) to strongly saline (8.1–16 dS m−1). In addition, the ECs was greater at a depth of 30–60 cm than it was at a depth of 0–30 cm. The ECs decreased from more than 8.0 dS m−1to less than 4.0 dS m−1after the first sampling period.