We have tried to determine whether statistical evidence on the occurrence of sympathetic flares, which is negative for whole-disk data, can be found for particular, physically connected, pairs of active regions. Recently, Simnett (1974) and Gergely and Erickson (1975) claimed to have found such evidence, but their results were based on incorrect computations of the random incidence of flares. If the correct formula is applied, the supposed evidence disappears. The results are negative also for pairs of active regions interconnected with magnetic loops visible in soft X-rays during the Skylab mission. The only positive result (with statistical confidence of 3.4 ~) is found for pairs of active regions, which are closer than 30 ~ to each other, without specifying any kind of physical relationship. For such pairs of regions the occurrence of short-time ( < 20 rain) intervals between flares is increased, but the time interval pattern does not correspond to any mode of propagation of a triggering agent in the solar atmosphere. Therefore, if the increase has real physical significance, it would be indicative of some kind of subphotospheric synchronization of activity in nearby 'sympathetic' active regions.