In traditional single-species fish stock assessment natural mortality is considered constant, independent of age or size. It is argued that predation mortality is generally related to prey size, that it decreases with increasing size and that such size-specific signals can best be detected in stocks that are close to a virgin state. Size-specific natural mortality rates of westcoast steenbras (Lithognathus aureti) were determined by using length frequencies of rod-caught fish from a lightly exploited and closed population at Meob Bay, Namibia. It was assumed that natural mortality is inversely proportional to (body) length and approaching a constant minimum rate (M_∞) as the fish grow bigger. Simple and new length-based catch curve methods were developed using the traditional simplification of neglecting variations in length at age. The von Bertalanffy growth parameters for annual mean growth as well as the coefficient of variation of length at age were estimated from analyzing age-length data. A simple deterministic simulation model was developed to examine the robustness of the methods and the impact of variability in individual growth. The model assumes that fish grow with a constant coefficient of variation in length at age. The simple method works within 10% precision criteria in most real cases. It is shown that overestimating mean length at old age (L̄∞) counteracts the effects of overlapping lengths for consecutive age groups. This fact can be used to estimate the fundamental mortality-to-growth ratio (M_∞/K) without any prior knowledge on growth. The application of M_∞ for steenbras to obtain size-specific natural mortality rate for silver kob (Argyrosomus inodorus), as an input parameter for virtual population analysis, is also proposed. Keywords: Natural mortality, Size specific rates, Length-based methods, Individual growth variation, Simulation, Lithognathus aureti.