We consider entanglement swapping for certain mixed states. We assume that the initial states have the same singlet fraction and show that the final state can have singlet fraction greater than the initial states. We also consider two quantum teleportations and show that entanglement swapping can increase teleportation fidelity. Finally, we show how this effect can be demonstrated with linear optics.PACS numbers: 03.67. Lx, 42.50.Dv Maximally entangled states are crucial for quantum information processing. If two parties share a pair of qubits in the maximally entangled state then they can perform quantum teleportation. However, usually the parties share nonmaximally entangled state ̺. One can define the singlet fraction F of the state ̺ as the maximal overlap of the state ̺ with the maximally entangled state, i.e.,where the maximum is taken over all maximally entangled states |Ψ . If one performs quantum teleportation with the state ̺ preprocessed by local unitary operations [1] then the optimal teleportation fidelity isIf F > 1 2 then the parties can perform quantum teleportation with the average fidelity of the teleported qubit exceeding classical limit 2 3 . However, it is well known that there are two-qubit entangled states which have F < 1 2 . The Horodecki family has proved that one can increase the singlet fraction of any two-qubit entangled state above 1 2 by non-trace preserving local operations and classical communication (LOCC) [2]. Verstraete and Verschelde have proved that one can do it even with trace preserving LOCC [3]. Moreover, they have found how to obtain optimal teleportation with any two-qubit state, i.e., how to find an LOCC protocol which gives the highest average fidelity of the teleported qubit. Let us now consider a string of nodes connected by nonmaximally entangled pairs of qubits. The first set of entangled pairs is distributed between nodes A and B, the second one is distributed between B and C and so on. In order to perform quantum teleportation from the first to the last node one first distills entanglement between A and B, then between B and C and so on [4]. Next, one performs entanglement swapping [5,6] at each node which creates entanglement between the first and the last node. Notice that usually many entangled pairs are needed between two nodes in order to distill entanglement. Finally, one performs quantum teleportation between the first and the last node. This strategy is crucial ingredient of quantum repeaters [7]. However, for pure nonmaximally entangled states there exists another strategy. Instead of distilling maximally entangled state between each pair of neighboring nodes and then performing entanglement swapping one first performs entanglement swapping and then distills entanglement between the first and the last node [8,9,10,11,12]. If each pair of neighboring nodes is connected by a single nonmaximally entangled pure state this strategy gives higher probability of obtaining maximally entangled state between the first and the last node.In this paper we conside...