The majority of mid-eclipse time determination has been determined by means of the Kwee-van Woerden method (KWM). There are also other possibilities e.g. to use physical models of eclipsing binaries or light curve templates and to determine mid-eclipse times using the least square method (LSM). We compared results yielded by both methods by means of a computer simulation on the synthetic model of ARAur primary minimum. Minima times determined by the KWM and the exact LSM approach are nearly the same, while the scatter of LSM times is always smaller than the scatter of KWM times. KWM uncertainties are systematically underestimated.