Context : A severe problem of the research in star-formation is that the masses of young stars are almost always estimated only from evolutionary tracks .
Since the tracks published by different groups differ , it is often only possible to give a rough estimate of the masses of young stars .
It is thus crucial to test and calibrate the tracks .
Up to now , only a few tests of the tracks could be carried out .
However , with the VLTI it is now possible to set constrains on the tracks by determining the masses of many young binary stars precisely .

Aims : In order to use the VLTI efficiently , a first step is to find suitable targets , which is the purpose of this work .
Given the distance of nearby star-forming regions , suitable VLTI targets are binaries with orbital periods between at least 50 days , and few years .
Although a number of surveys for detecting spectroscopic binaries have been carried out , most of the binaries found so far have periods which are too short .

Methods : We thus surveyed the Chamaeleon , Corona Australis , Lupus , Sco-Cen , \rho Ophiuci star-forming regions in order to search for spectroscopic binaries with periods longer than 50 days , which are suitable for the VLTI observations .

Results : As a result of the 8 years campaign we discovered 8 binaries with orbital periods longer than 50 days .
Amongst the newly discovered long period binaries is CS Cha , which is one of the few classical T Tauri stars with a circumbinary disk .
The survey is limited to objects with masses higher than 0.1 to 0.2 M _ { \odot } for periods between 1 and 8 years .

Conclusions : We find that the frequency of binaries with orbital periods \leq 3000 days is of 20 \pm 5 \% .
The frequency of long and short period pre-main sequence spectroscopic binaries is about the same as for stars in the solar neighbourhood .
In total 14 young binaries are now known which are suitable for mass determination with the VLTI .