Plant survival was evaluated in a statistical analysis. The analysis was conducted based on data from a long-term field experiment where five willow (Salix L.) genotypes were grown at four densities (12,000, 24,000, 48,000, and 96,000 ha−1) in a plantation with an annual harvesting cycle. Four exponential power distributions that best fit plant survival data were selected, including an alternative parametric distribution, which is presented for the first time in this study. The developed models supported predictions of the future distribution of willow plants in groups characterized by different planting density, willow cultivars, and clones. At the recommended density of 24,000 plants ha−1, the exponential power distribution proposed by Smith and Bain best fit willow survival time and the plantation’s productive lifespan. The new hyperbolic secant distribution was characterized by the highest explanatory power measured by Akaike information criterion, Schwarz’s Bayesian information criterion, and a heavy-tailed distribution. This model was optimal for higher planting densities (48,000 and 96,000 ha−1) with significantly lower plant survival rates.