Effusion is one of the most elementary kinetic processes, to which the concepts of nonequilibrium statistical mechanics can be applied. Remarkably, the stationary probability distribution can be exactly constructed as a so-called Poisson suspension, explicitly showing that time reversal is broken at the statistical level of description under nonequilibrium conditions. The multivariate fluctuation relation for the energy and particle currents can be directly deduced from the underlying microscopic dynamics. Moreover, temporal disorder and its nonequilibrium time asymmetry can be fully characterized and shown to be related to the thermodynamic entropy production. The multivariate fluctuation relation can also be applied to mass separation by effusion.