Due to the increasing number of multilingual students at Austrian and German schools numerous concepts of additive and integrated second language teaching and learning have been developed during the last decade. Irrespective of their structures these concepts are effective only if teachers know about their students’ language skills. For this purpose, several language assessment tools were constructed to identify students in need of additional language lessons or to support in first and second language acquisition within their school careers. Apart from tests analysing tools like profile analysis and analytical observations are relevant in the context of teaching and learning at school. These tools allow precise qualitative descriptions of multilinguals’ language competence an offer an extensive basis for didactical decisions. I intend to demonstrate the potentials and challenges of analytical observations, which refer to Pienemann's Teachability Hypothesis and condense results of language acquisition research. For this purpose I will report on structural models, validity, reliability and objectivity of the two observation tools Niveaubeschreibungen DaZ (Language level descriptors for German as a second language) and Unterrichtsbegleitende Sprachstandsbeobachtung DaZ in Österreich (Lesson immanent observation for German as a second language in Austria) and point out requirements and consequences for teacher training.