This paper concerns sensor network key distribution schemes (KDS) based on symmetric-key techniques. We analyze the problem of active attacks against such schemes. By active attacks we mean those attacks, where the adversary can maliciously disturb the communication between the sensors. We observe that the active adversary that captured even a small number of sensors, can anyway get a full control over the network, no matter how strong the KDS is. Therefore we conclude that the best scheme in this context is the one based on the method of Bl¨om (1984) (which guarantees perfect secrecy of the keys, as long as the number of corrupted sensors is small).