The German logistics industry continues to gain importance as the population increases. At the moment more than 70% of goods are transported on the road. From 2011 to 2014 volumes in road transport increased by 2.5%. By 2030, the increase of the transport volume is expected to be 16.8%. Compared to total tons transported by rail, inland waterways, sea, air and oil pipeline, the road transport volumes are more than ten times. The increasing importance of logistics and its effects on transportation processes have been considered in freight transport demand modelling by new model developments incorporating several logistics aspects. However, transport logistics hubs seem to be not considered adequately in most freight models although they are crucial for freight transport. In order to depict the future developments of road freight transport, models for freight transport are very important as the traffic volume is continuously increasing. To come up with such models, patterns in the behaviour of German logistics service companies must be studied at first. For instance, the choice of locations provides information about the formation of transport routes while other attributes that characterize the locations, can give a better view about traffic flows and traffic congestion. Additionally, the corporate types of logistics service companies also need to be taken into account, in order to study the structure of those companies. Therefore, the aim of this work is the development of a classification method for different logistics locations. This leads to a better understanding of the purpose for each location. In subsequent work, this classification can be used, for example for the disposal of group-specific density rates. First of all, for a classification of logistics sites the number of sites that can be investigated is required. As an official list of companies operating in the logistics sector does not exist, the total number of logistics companies in Germany is currently unknown. This makes the conduction of freight demand models with aspects of typological strategies of logistics companies arduous. Only by means of empirical surveys an exploration of this area is possible, because there is no directory from which samples for the survey can be drawn from. Therefore, a cost-saving analysis of logistics locations based on secondary data, helps to gain new insights. But before locations can be gathered, characteristical features for the sites are needed, in order to enable a structured and uniform data collection. These features contain specific logistics- (e.g. hall layouts, amount of pallet space), traffic- (e.g. amount of ramps and vehicles) and site-related information (e.g. number of employees, generated revenues). A literature review on the existing works provides a number of features which describe the locations best. In addition to these features, the standard location types are defined. Once completed, a set of distinctive figures of site types are depicted in standard location profiles. To exclude the ambiguity of definitions, a conceptual framework of all important attributes for the analysis of site types takes place in the first part of the work. After the frame for the data collection is defined, German logistics hubs can be recorded. These sites are implemented into a Microsoft-Access-based location database. Having an address is the minimum requirement for the import of the sites into the database, but gathering the defined features must not be neglected as they are needed for the assignment of the standard location types. Subsequently, the location data base contains a large amount of locations, which are partly filled with characteristic attributes. But how can meaningful results be drawn out of this large dataset? This question is going to be answered in the next step. Here, statistical methods, which compress the information and make them comparable, need to be chosen. This descriptive statistical analysis includes the application of statistical parameters, charts, graphs and tables in order to describe characteristic parameters of the sample. There are three different types of statistical characteristics. A distinction is made between qualitative, ordinal and quantitative characteristics. These types of features are processed depending on the existing scale level. Location parameters such as median, mode or arithmetic mean inform about the central distribution of the characteristics. Dispersion parameters such as standard deviation, variance, or coefficient of variation are used to assess the variance around the average value. Thus, by using these parameters, it is possible to mark outliers. Furthermore, these parameters also make the site types comparable and via the comparisons with self-collected data, discrepancies can be detected. The first part of the empirical analysis includes the application of statistical indicators at all locations with assigned site types, yet for some types, some needed information is absent. For this reason, in the second empirical part all locations of the database according to the types of companies are interpreted, in which market reports, industry reports and analysis of the submarkets were taken on hand to reconcile the data on enterprise level. As the total population of logistics companies in Germany is unknown, statistical methods have to be used. With inductive statistical ratios, it is possible to extrapolate the location and dispersion parameters on a large amount of data and should be used as estimations of possible distributions. Whether this is consistent with actual values or not has yet to be proved. Therefore, data has to be selected primarily. The reports of the sub-markets and the analysis of the data demonstrate that a representative proportion of locations is still not fulfilled in all areas. In the last section of the empirical elaboration, a selected structures-discovery-process application is used. These methods include classification methods, which summarize the objects based on the defined characteristics. To make sites classifiable, qualitative and quantitative criteria, which are explained in the first theoretical part, are again required. Several classification methods have been explained theoretically and tested for use on the locations of the logistics location database. Among these methods, the Two-Step-Cluster-analysis is considered useful to categorize objects, since this classification serves to slim down the amount of data. By noise selecting, it is also possible to make out outliers and exclude them from the evaluation. The aim of the Two-Step-Cluster-analysis is to recognize typical types of location within the logistics subareas. As a result, for each site type, companies with typical characteristics can be found and compared with own collected data. The Clustering depends on the defined characteristics, such as area, number of vehicles or ramps, as well as the location type. 17 combinations of features for the cluster analysis are formed. First, these combinations are used on all locations with a deposited site type. Afterwards a differentiation, according to the corporate type, is made. So the cluster combination is calculated for the six corporate types: “third-party logistics provider”, “CEP-provider”, “third-party logistics provider specialized on groceries”, “transhipment service provider”, “transport service provider” and “third-party logistics provider specialized on beverages”. Now, two information can be gathered out of the date. The first is the affiliation of site types to clusters. If different site types are collected in one cluster, similarities between these sites can be investigated. Also, the differences between site types, which occur in other clusters, can be seen and delved into. As the second information, for each combination of features the assignment of sites can be found. If the same sites are used in different cluster combinations, these sites have many assigned features. To conduct primary empirical surveys, the locations are selected, which are processed in the individual clusters and occur in different clusters. Therefore, the classification methods help to reduce data. Additionally, these companies can be selected for traffic surveys. Comparison between secondary and primary data will be possible by using the sites shown in the different clusters. The main results of this thesis is therefore the selection of appropriate companies and logistic hubs. Clustering provides possible approaches for choosing appropriate locations. Now, it is possible to find out the amount of freight transport produced by those sites. For freight transport and land use planning policies these insights are highly relevant, as there is no model in Germany that gives information about the increasing traffic volumes for each site type. So the foundation for freight transport demand modelling approaches are set. Furthermore, with these freight transport modelling approaches optimal spots for sites can be suggested. This is not only beneficial for the companies, which may have better and cost-saving routes for their transports, but also negative effects, like noise and air pollution can be reduced. Here it may also be interesting whether it is possible to lower the amount of generated traffic, by rearranging the sites and putting them on places which is best fitting for their economical purposes.