Ejector-based refrigeration cycles working with natural refrigerants have already gained industry attention. Replacing throttling valve with an ejector in vapour-compression cycles brings high improvement of the cycle efficiency due to the ejectors potential of recovering part of throttling losses. With the rapidly growing market for heat pumps, they are also being implemented in those cycles, but this requires redesigning the ejector geometry for new natural working fluids for different operating conditions and applications. Typical approach to perform the ejector shape optimization is to use the ejector mass entrainment ratio or overall efficiency as an objective function. However, an entropy generation analysis seems to be more efficient. For this reason, the aim of this work was to perform the numerical analysis of the entropy generation of the two-phase ejector for R744 and assess its potential as a tool for efficiency improvement in the shape optimization algorithms. The ejector PL tool utilizing the homogeneous equilibrium model approach was complemented with the entropy generation model implemented using an additional transport equation to the computational fluid dynamics software. The numerical results of the mass flow rates were used for validation purposes. The entropy generation module allowed for the entropy generation analysis in terms of maximum values and their location showing critical areas of irreversibility characterizing different working fluids usage and ejector applications.