The issues of multilingual access to legal information are examined, and strategies of cross-language retrieval to legal information resources are illustrated as additional services of the Portal to legal literature created by ITTIG. Consideration is given to the peculiarities of legal language as a technical language closely related to the diverse legal systems. The paper describes a methodological approach planned for the Portal to provide a single point of access into disparate repositories where categories of law (i.e. trade law, constitutional law, criminal law) are the essential metadata to point to relevant material irrespective of the language used in a query. Categories of law of a specific legal system represent the way how retrieval can be satisfactorily achieved. Strategies and techniques for translating legal queries to different target languages, eventually disambiguating ambiguous words by a machine learning approach are illustrated.