The paper describes the study carried out in the area of Colosseum, Roman Forum and Palatine Hill in Rome, selected for its unique characteristics and importance, being the Italian most visited touristic site. Data on the sonic environment and appraisal of some environmental features, including a few non acoustic ones (like landscape and smell), have been collected in 8 sites selected to cover different characteristics. The acoustic data have been determined from binaural recordings taken at the above sites simultaneously with subjective ratings collected by a structured questionnaire filled in by 212 interviewees. Statistical analyses have been applied to extract main features and to classify the sites by the unsupervised learning algorithm of k-means clustering. Among the various solutions, the one that groups the data into two categories looked the most appropriate: in one the sonic environment is dominated by technological sources, whereas in the other one anthropic sources are predominant.