All of the sampling plans currently in use for general population surveys in Germany suffer from methodological and practical problems. A new sampling plan is thus urgently needed: one with a low cost overhead that can be prepared in a very short time. Germany also lacks a sampling plan covering all institutional populations, immigrants in general, and illegal immigrants in particular. The availability of new databases covering these populations suggests ways of developing, implementing, and testing new sampling plans for population surveys in Germany. One such sampling plan (G-Plan) is proposed here for the first time. The implementation problems of this design must be studied in a number of empirical pretests.