Currently, the most popular approach for assessing the tsunami hazard on a coast is the PTHA (Probabilistic Tsunami Hazard Assessment). In this study, we develop one of the variants of the SPTHA (Seismic PTHA) method, adapted to solving near-field tsunami zoning problems. The approach is applied to assessing the tsunami hazard of the Bechevinskaya Cove located on the eastern coast of the Kamchatka Peninsula in the northern part of Avachinsky Bay. We propose the method, algorithms and results of probabilistic assessment of the cove’s tsunami hazard in order to determine the safest water areas, in which the values of the intensity measures (IMs) of tsunami will not exceed the specified threshold values with the given recurrence rates. The method includes analysis of seismotectonics of the region, construction of a catalog of model tsunamigenic earthquakes, determination of their statistical characteristics, scenario numerical modeling of the dynamics of tsunami waves, calculations of the values of IMs that can be exceeded with the given recurrence rates (on average 1 time in 100, 500, 1000 years). Spatial distributions of the maximum wave heights and maximum velocities are provided for the recurrence rates. Three configurations of the water area are considered, including the possibility of constructing protective structures, and conclusions are drawn about their influence on the tsunami hazard assessments in the cove. 1. Introduction The problem of tsunami zoning, which consists in the quantitative classification of the coast and adjacent aquatory by the degree of tsunami hazard, is one of the classic tsunami problems and requires the use of multidisciplinary tools and methods for its solution (Gusiakov, 2017). Hereinafter, tsunami hazard is the degree (level) of exposure of water areas and coasts to the tsunami threat in terms of measuring the catastrophic wave intensity. One of the first approaches to solving the problem of tsunami zoning was a method based on the use of historical data, materials from modern instrumental observations, as well as geological studies of paleo-tsunami along the coastal areas. The main limitation of the “historical” approach is a short series of observations. For near-field tsunami zoning problems the number of such observations is usually objectively limited by the short historical period of observations and the preservation of tsunami deposits, which also limits the accuracy of statistical estimates of tsunami hazard. Scenario modeling is an alternative based on estimates of the statistical properties of tsunami sources and the results of simulation (mathematical modeling) of the transformation of tsunami waves from the source to the coast. This approach uses a series of model tsunamigenic events of arbitrary length and thus clarifies probabilistic estimates of tsunami hazard. One of the ways to use the results of scenario calculations to solve the problem of tsunami zoning is to determine the extreme (worst case)