The task to estimate the fitness of a clonal plant with polyvariant ontogeny reduces to compiling a life cycle graph, constructing and calibrating the corresponding matrix model of the discrete-structured population, and calculating the dominant eigenvalue (λ1) of the model matrix. We demonstrate a solution to this task with a sample of Calamagrostis epigeios , a perennial long-rhizome grass propagating vegetatively, and data on the age-stage structure of its local population. A traditional technique of successive censuses fixing the age-stage status of all individuals on a permanent sample plot (SP) provides for calculating frequencies of ontogenetic transitions directly from the data, but leaves uncertain the status-specific reproduction rates as the recruit parents are unknown (“reproductive uncertainty”). Uncertainty in data was leading to that in the estimation and dictating a need to change the method of field study: the description of above-ground parts of plants has been completed with the analyses of rhizome parent-daughter links revealed after digging SPs out. However, the traditional, fixed area of SPs (1 m 2) forced cutting the links off along its perimeter, while those within the SP turned out quite entangled already in a 4-year-old colony. A result, the reproductive uncertainty were not eliminated completely, and the next step in the method development has become to determine the contour of the entire woodreed colony and to carefully dig it out. Analysing both the above- and below-ground spheres of the colony has enabled us to calculate uniquely all the elements of the matrix model, hence the value of λ1, while accounting for the actual area of the contour in the current and previous years amends the value of λ1 needed for comparison with the results of previous studies.