A total of 6400 noise measurements were taken in downtown Medellin, a 2km x 2km area to characterize noise levels. The area was divided into a grid of 160 blocks with a length 154 m each; 40 measurements were taken in each block. Sampling was conducted for 10 days, with simultaneous measurements in four sampling zones selected randomly and with coverage of four sampling periods (7:00 - 9:00 AM, 9:30 -11:30 AM, 12:00 AM - 2:00 PM, y 5:00 - 7:00 PM). The zone between 44,h and 49,h streets and between 40,h and 49th avenues has the maximum average noise level with 74 dBA (decibles in scale A). Four of the noisiest sites were measured during 48 hours continuously by usage of a dosimeter. Auto correlation results allowed use of Geostatistics to built isophones for the four time periods. Isophones plots overimposed on physical layout of the city shows a good correlation between high noise levels and heavy traffic through main avenues (Avenida Oriental, Calle San Juan, Avenida del Ferrocarril, Bazar de los Puentes, Glorieta de la Minorista). Medellin is a noisy city that does not meet existing national regulations. The proposed methodology based on Geostatistics and experimental design is a feasible and comprehensive approach to manage urban noise.