Knowledge of the population dynamics of tropical trees has expanded considerably in the past 20 years. An important observation deriving from these investigations is the confirmation that population behaviour varies both in time and space. A tool recently developed to evaluate the potential for variation in vital rates, and therefore in population growth rate, is variance-standardized perturbation analysis (VSPA). In this paper we report the results of a 2-y demographic analysis of a population of the tropical tree Manilkara zapota in a subtropical rain forest in the Mexican state of Veracruz, in which variance-standardized perturbation analysis was applied and compared with the results of the traditional elasticity analyses. To build population projection matrices, we tagged and followed a sample of 91 juvenile and adult individuals, and 635 seedlings. We subdivided the sample in nine size classes (defined by tree height and dbh; as well as leaf size, in the case of seedlings) and estimated transition probabilities and fecundity for each class. The demography of M. zapota varied greatly from the first to the second year of study (in 1998–1999, λ = 0.987, while in 1999–2000, λ = 1.038) due to negligible seed production during the first year and a massive reproductive event during the second. The largest elasticity values for both years corresponded to persistence of large juveniles and adults. Although the fecundity entries showed very low elasticity values, the variance-standardized perturbation analysis revealed the importance of these matrix entries; transition to larger categories and retrogression to smaller ones of saplings and juveniles were also important demographic processes contributing to variation in λ according to the VSPA. Thus, although the results of elasticity analysis and VSPA were similar for the 1998–1999 matrix, they differed substantially for the 1999–2000 matrix. In the latter, the VSPA enhanced the importance of demographic processes that are intuitively relevant for the population studied. This points toward the necessity of further exploring the use of VSPA, since it offers several advantages over the traditional elasticity analysis: it concentrates on the impact on λ of vital rates that actually vary, and the interpretation of the results is more realistic and straightforward.