The task of changing the overlap between two quantum states cannot be performed by making use of unitary evolution only. However, by means of a unitary-reduction process, it could be modified. Here we study in detail the problem of mapping two known pure states onto two other states in such a way that the overlap between the outcome states is different from the one of the initial states. We show that the modulus of the overlap can be reduced only probabilistically, whereas it can be increased deterministically. Our analysis shows that the phases of the involved overlaps play an important role in the increase of the success probability of the desired process.The unambiguous state discrimination processes [1-9] can be viewed as probabilistic conclusive mappings connecting a set of initial states with nonvanishing overlap with a set of orthogonal final states. One can think in terms of a more general process consisting of a probabilistic conclusive mapping which connects two sets corresponding to initial and final states, each of them with different overlaps. The probabilistic cloning scheme is a particular example of this [10]. Recently, the mapping between sets of nonorthogonal states has been connected to the control of quantum state preparation, entanglement modification [11], and interference in the quantum eraser [12]. The role of majorization in describing quantum operations for probabilistic and deterministic dynamics of a system has been studied with respect to finding the conditions for mixing enhancement for all initial states [13]. Zhou et al. constructed the unitary implementation of state transformations in a composed system, obtaining the necessary and sufficient conditions for the existence of the desired maps [14].In this work, we study the conclusive mapping between two pairs of states with different overlaps. The transformation is performed by means of a unitary-reduction process on the primary and an ancillary system. We allow different a priori probabilities of the two initial states, and we take into account the complex nature of the overlap between the states. We find conditions that guarantee the existence of the mapping, and we obtain the optimal success probability. Also, we discuss the relation of the proposed mapping with some known quantum information protocols, such as quantum deleting and quantum cloning. We compare the optimal success probability of our scheme with the one obtained via an alternative scheme using the unambiguous state discrimination.Let us consider a system s described by a two-dimensional Hilbert space H s . This system is prepared randomly in one of the states |α i s (i = 1, 2), with a priori probabilities η i . Our aim is to map the states {|α i s } onto the states {|β i s }, where the overlaps α 1 | α 2 = α and β 1 | β 2 = β are different. The mapping must be implemented conclusively, that is, it must be known with certainty when it has been perfectly carried out, and with the highest possible success probability.In order to implement the mapping, we introduce an ...