In classical physics the joint probability of a number of individually rare independent events is given by the Poisson distribution. It describes, for example, unidirectional transfer of population between the densely and sparsely populated states of a classical two-state system. We derive a quantum version of the law for a large number of non-interacting systems (particles) obeying BoseEinstein statistics. The classical low is significantly modified by quantum interference, which allows, among other effects, for the counter flow of particles back into the densely populated state. Suggested observation of this classically forbidden counter flow effect can be achieved with modern laser-based techniques used for manipulating and trapping of cold atoms.PACS numbers: 03.65.-w, 03.75.lm, 02.50.-r In classical physics and statistics, probability for a number of individually rare events is universally given by the Poisson distribution (see, for instance, [1]). For example, it is obeyed by a classical gas escaping into an empty space through a penetrable membrane. With the number of atoms N large, and the transition probability made proportionally small, the number of escaped atoms is governed by the Poisson law, with the number of atoms recaptured into the original reservoir vanishing as N → ∞. The validity of the Poisson distribution depends on that one can, in principle, know not only how many but also which of the atoms have escaped. Quantum mechanics offers a different possibility: for identical particles one is allowed to know only the number of the escapees, and not their identities. While it is well known that both Fermi-Dirac and Bose-Einstein symmetries of a wave function may lead to non-poissonian effects in the full counting statistics of otherwise independent particles [2]-[8], the failure of the Poisson law in the limit of rare events is less obvious. The subject of this Letter is the general question of what replaces the classical Poisson law in a quantum situation where only the total number of rare events, but not their individual details, can be observed?. We specify to the case of many non-interacting bosons, each of which may occupy one of the two available states. Such systems are also of practical interest, e.g., for their potential applications as detectors. For example, if the transmission amplitude between two connected cavities is influenced by a passing particle, the change observed in the photonic current would announce the particle's arrival. In a similar way, atomic current of a weakly interacting Bose-Einstein condensate (BEC) trapped in a double-or multi-well potential (see Fig.1) can be used to gain information about the state of a qubit coupled to the BEC [9]- [11]. Detailed analysis of the work of such hybrid bosonic devices must take into account in which manner, and how frequently, the bosonic sub-system is observed, and will be given elsewhere. We note that the problem is fundamentally different from that of the frequently studied coined quantum walk [12], where interference betw...