A forecast of future occurrences of larger (M ≧ 5.9) crustal earthquakes in Italy is made on the basis of historic and instrumental seismicity, and following a seismotectonic regionalization recently derived on the basis of geological evidence combined with earthquake epicenters and focal mechanisms. Completeness problems in the seismic catalog allow one to study only 20 regions out of 58. Large crustal seismicity within each seismogenic region is modeled as either a periodic Gauss process or a random Poisson process according to the experimental coefficient of variation of the series of past occurrences in each region. Return times are estimated directly from the series of (M ≧ 4.5) earthquakes if they are sufficient in number, or from the Gutenberg-Richter law applied to lower magnitude seismicity (M ≧ 4.5) otherwise. The immediate probability of an M ≧ 5.9 crustal seismic event is estimated to be very low in all regions except southeastern Sicily and Appennino Abruzzese. In the near future (next 20 yr), the estimated probability is high (above 65%) also in the Appennino Forlivese and Naso-Capo d'Orlando regions. In addition to detailed seismic risk reevaluations, these regions represent the best bet for a program of intensive monitoring to gather a record of the process of strain accumulation and seismic release.