RESUMO
The objective of this study was to predict chloroform formation resulting from the process of disinfecting water, particularly trihalomethane which is most frequently produced. A statistical model was used which included repeated measurements of water parameters used for monitoring water quality at 51 sites covering the municipal water system of Montevideo. Samples were taken considering different seasons from June 2009 to July 2011 in Montevideo. Total samples (n = 330) were analytically studied using the headspace-gas chromatography method coupled with mass spectrometry. Chloroform was the dependent variable and the covariables were pH, temperature, free chlorine, and total chlorine. A Tobit analysis with an unstructured correlation matrix was performed, and a significant interaction was found between pH and free chlorine for the prediction of chloroform formation. We concluded that parameters for the continuous control of water quality for consumption can be used to predict the levels of chloroform that may be present. Given the large measurement to variability found in the repeated measurements, the use of averages that include more than one season is not recommended to determine the degree of compliance with acceptable levels established by norms.