We are now at the end of a long symposium which lasted five (but not all full) days. The objective was first to review the best present achievements of astrometry and to present projects for a much more advanced astrometry in terms of precision attained versus the number of stars and limiting magnitudes. The word sub-milliarcsecond astrometry was intentionaly vague. It has rightly been often understood as designating a range of precisions of the order of a few tens microarcseconds.