I have a table 1 of 10000 records with the following colums:
Prak (identification practice), DNr (identification patient), ************, YearOfBirth, DateDiagnose
and a table 2 of 135000 different records with:
Prak (identification practice), DNr (identification patient), ************, YearOfBirth

Now I want to ad the date of diagnoses of t1 to the patients in t2 with the same ************ and YearOfBirth

The problem is that if I link ************ and YearOfBirth the tables multiply.

This is not enough information. You need to tell us what fields are unique if any. I think what you need to do is link off of a combination of [Prak (identification practice) and DNr (identification patient)]. ************ and YearofBirth is not unique by patient so you can't link off them without risking a cartesian product(bad join). You need to find unique data between the tables to link off of, like an ID field of some sort.

Table 1:
Prak DosNr Gesl GebJr DatDiagn
M0101 7 F 1923 13/12/1997
M0101 25 F 1923 16/07/1995
M0101 123 F 1923 1/11/2003
M0102 15 M 1923 15/7/2000


Table2
Prak DosNr Gesl GebJr
M0101 223 F 1923
M0101 1013 F 1923

I want to give the patients of table 2 give ad random a date of diagnosis of a patient with a patient of the same ************ and year of birth from table 1.
Patients of the same ************ and year of birth are 'matched' for statisrtical reasons.
It is possible to give for example the MIN or MAX date to patients with the same ************ and year of birth, but the different dates had to been seen in table 2.

I don't think you can link off a "min" or "max" directly in SQL but you could select min or max into a temp table and link off that if I understand you right.

That's possible, but then all the patients in table 2 with the same ************ and year of birth have the same date of diagnosis. I try to give them the different dates of table 1.
I think it's not possible by a direct link.