In this paper, we evaluate the spatial location patterns of Spanish manufacturing firms and we assess the different tendencies to cluster in each industry. To do this, we use a distance-based method Puech, 2003, Duranton andOverman, 2005), more concretely the Ripley's K function, which allows us to treat space as continuous, measuring concentration by counting each firm's average number of neighbours within a circle of a given radius. In this way, K(r) function is used to describe characteristics of the point patterns at different geographical scales at the same time, letting us detect the statistical significance of departures from randomness.Our approach incorporates an additional improvement. By means of the software employed, 'R', we can apply border corrections adequately in any irregular polygonal shape; therefore, we introduce a polygonal envelope, which allows us to improve the delimitation of our area of study, Spain, avoiding the nuisance of empty spaces.We apply this method to Spanish manufacturing sectors at two-digit level and we realise that results depend on the benchmark employed. In fact, whether we use 'complete spatial randomness' as benchmark, every sector analysed presents significant concentration whatever the distance of the radius we consider. However, as is well known, this type of analysis is sensitive to the benchmark considered, thus we also use as benchmark the whole of manufacturing sectors. In this case, we find patterns of localization completely different to the previous ones; appearing dispersion in some sectors relative to all manufacturing firms. Therefore, by means of this method we can know characteristic features of the pattern of localization of every manufacturing sector, like whether concentration or dispersion exists, which is its intensity and at which distance, or geographical scale, we obtain its highest level.