OBJECTIVES: The survey was initiated to gain insights into the implementation of decision-making tools in national guidelines in treatment patterns of mCRC in daily practice. The tools define treatment intensity for subgroups of patients by clinical characteristics and anticipated treatment aims. METHODS: Physicians in representative sample of centres (69) reported all pts. with treatment decision in October-December 2009. The database contains 1019 pts. with retrospective record of treatment history. Treatment decisions were analysed in 3 predefined subgroups according to a model based on the German guidelines.  Statistics were performed in SPSS by bivariate analyses with two-sided Chi-square test. In the next step the subgroups of patients with comparable treatment in clinical practice were defined and analysed in a multivariate analysis. RESULTS: By contrast to the recommendation of national guidelines, intensified therapy was administered less frequently in patients with the aim of “resection of metastases”  (43%), whereas the highest use (64%) was reported in “patients with tumor related symptoms or at risk for rapid progression or deterioration”. This group represents only 12% of the 1st line. Factors with an influence on the use of intensified therapy in daily practice were analyzed in a multivariate analysis. Three treatment clusters (comprising 89% of patient sample) were determined. (all p<0,05) The cluster with significantly higher use of intensified therapy (+18% above mean value of 54%) is distinguished by: Age <70 y., better PS (>=80% KI), no symptoms and/or without concomitant diseases, treatment in office based setting. Patients in this cluster show less tumor dynamics. CONCLUSIONS: In daily practice, the application of the decision model based on treatment aims for clinical subgroups is not generally used. Intensified treatment is more likely associated with individual patient characteristics and institutional framework. This, however, underlines the need for a critical discussion of the currently suggested decision-making models.