The problem of noise levels in the home is of increasing interest for reasons of health and psychological well-being, among others. This study analyses in detail the meta-data compiled in the Spanish Census of Population and Housing, which provided data on a large number of environmental variables, including noise pollution in the home. A geostatistical study is conducted from the data provided by the latter survey, in which spatial autocorrelation is measured by analytical techniques by using R statistical software and the ESRI geographic information system. We study the empirical variogram, analize different theoretical models (gaussian, exponential and spherical) and estimate the parameters from different perspectives: constant or linear trend, weighted least squares, maximum likelihood and restricted maximum likelihood. The noise level is estimated through the kriging interpolation technique, using the different parameters of each model.