Recently, an attractive link-selection scheme has been proposed for nonregenerative relaying networks, by considering one source, one relay and one destination. In that scheme, based on a distributed mechanism, either the direct link or the relaying link is preselected at the transmitter side, before each communication process. This saves not only feedback overhead, when compared with centralised mechanisms, but also time slots, when compared with fixed relaying along with selection combining at the receiver side. Herein, we extend the design and analysis of such a scheme in order to allow for multiple relays and multiple destinations. In such case, before the link selection mechanism takes place, a destination and a candidate relay must be chosen. We consider two different relay-selection policies: the locally optimal one and the so-called max-min. The analysis is provided in terms of outage performance and spectral efficiency. As in related studies, an exact treatment proves intractable, because the underlying probabilities are extremely intricate. Instead, for the locally optimal scheme, we provide integral-form lower and upper bounds for the two performance metrics, as well as closedform asymptotic expressions for each outage bound. For the max-min scheme, we provide a closed-form lower bound and an integral-form upper bound for the outage probability, as well as closed-form asymptotic expressions for each of these bounds and an exact integral-form expression for the spectral efficiency. Interestingly, our results show that the proposed distributed link-selection scheme is as good as its centralised counterpart and barely affected by the choice of the associated relay-selection policy.