Adsorption of natural organic matter by physically and chemiclally modified activated carbons was studied in this work. Principal component analysis (PCA) was applied to datasets gathering morphological, physical, chemical properties, and adsorption capacity information of the modified activated carbon samples. The PCA was used to identify the main effects by clustering the samples according to the modifications undergone. Results indicate that total and/or fractional pore volume, and the iron and oxygen content were sensitive enough to detect good adsorption potential even for "a priori" nonoptimal overall properties. Unexpectedly, low values of pHpzc not always correlated with poor adsorption performance. The sum of N and O content may also be a good indicator to select a carbon with good adsorption performance. Based on PCA results, a prediction equation was developed to estimate values for the equilibrium uptake capacity and was tested against five additional carbon samples.