Context :

Aims : The observation of eclipses during X-rays flares taking place in active cool stars binaries allows us to calculate the position and size of the flares .
This information can not be derived by analyzing the decay of the flares , a frequently used approach in the literature that requires the assumption of a physical model .
We make use of the eclipsing light curve to constrain the set of possible solutions , from the geometrical point of view , in two flares of Algol , and one flare in VW Cep .

Methods : We make use of a technique developed with the system SV Cam ( i \sim 90 \degr ) and generalize it to binary systems with arbitrary inclination .
The method simulates all possible geometrical situations that can produce the times of the four contacts of the eclipse .
As an approximation we assume that the emitting region has a spherical shape that remains unchanged during the eclipse .
We however show that this is a good approximation for the problem .

Results : The solutions observed indicate that in two of the three cases the flare can not be polar ( | \theta| < 55 \degr ) and in a third one the flare can be placed either near the pole or at other latitudes .
The emitting regions must have a small size ( 0.002 - 0.5 R _ { * } ) , but if interpreted as the apex of coronal loops , their length could actually be up to 3.1 R _ { * } for one of the Algol flares .
These measurements imply a lower limit to the electron density in the emitting region between \log n _ { e } ( cm ^ { -3 } ) 10.4 and 14.0 , and a magnetic field between 70 and 3500 G. Similar results are found if the emitting region is assumed to be loop-shaped .

Conclusions :