Abstract. Occurrence frequencies, OF, of 12-h precipitation amounts, P, at stations in the former European USSR are displayed as dependent on dynamic forcing of vertical motions. The dynamic forcing is described by a "frontal parameter", F (calculated in the points of objective analysis grid), which depends on the surface pressure field curvature and on the baroclinicity in the lower half of the troposphere. The precipitation rate spectra for 4 seasons, calculated from a large sample of data (7 years, about 650 000 values of P for one season), show a monotonous OF growth of all ranges of P>1 mm/12 h with F increase. The growth is especially significant for heavy precipitation. As a result, F is shown to be an informative predictor of P spectrum or of probability of any given range of P. As a next step, two-dimensional spectra of precipitation occurrence frequency, as a function of F and LNB, that is, OF (F, LNB), are calculated, LNB being the level of neutral buoyancy at the gridpoint, an estimate of grid-scale convective instability. On this basis, an approach to probabilistic forecasting is suggested.