OBJECTIVES Although subgroup analysis in clinical trials is often criticized, it is still considered an important part of the AMNOG Dossier when describing effect modification in different patient groups. An identification of effect modification in the subgroups will either support or weaken the total additional benefit (Zusatznutzen) of a drug. The purpose of this article is to give a detailed background with regard to the statistical inference in subgroup analysis and a brief review of the effect of tests for homogeneity/interactions on the final grading of the additional benefit, according to the decision from IQWiG. METHODS This article covers: Understanding tests for homogeneity, individual versus pooled data and influence of subgroup analysis on recent IQWiG benefit assessment. A research of the recently published AMNOG Dossiers was performed. A description was given to the number of drugs that had an additional benefit and the subgroups that were involved. Clustering analysis was performed to investigate the hidden structrue in the subgroups. Regression models were used to analyze the relationship of the subgroups and the additional benefit.  RESULTS Some subgroups such as age, gender and weight play a major role in the AMNOG assessment. Other subgroubs are more specific for certain disease areas e.g. BMI or hypertension for diabetis.  In a worst case the benefit of a drug may disappeare when a certain subgroup is taken into consideration. CONCLUSIONS Subgroups exert a profound influence upon the overall effect of a drug. Despite of the weakness of the statistical inference, subgroup analysis plays an important role in the AMNOG Dossier. Some subgroups are frequently related to the overall effect of a drug and some might even change the whole story in the AMNOG Dossier. In conclusion, subgroup analysis should be understood properly and the results should be interpreted carefully.