Object-based stochastic modelling techniques are routinely employed to generate multiple realisations of the spatial distribution of sediment properties in settings where data density is insufficient to construct a unique deterministic facies architecture model. Challenge is to limit the wide range of possible outcomes of the stochastic model. Ideally, this is done by direct validation with the 'real-world' sediment distribution. In a reservoir setting this is impossible because of the limited data density in the wide-spaced wells. In this paper this uncertainty is overcome by using size, shape and facies distributions of tidal channel and tidal flat sand bodies in a highly data-constrained lithofacies architecture model as input for the object-based stochastic model. The lithofacies architecture model was constructed from a densely perforated (Cone Penetration Tests and cored boreholes) tidal estuarine succession of the Holocene Holland Tidal Basin in the Netherlands. The sensitivity of the stochastic model to the input parameters was analysed with the use of varying tidal channel width and thickness values and calculating the connected sand volume per well for the different scenarios. The results indicated that for a small well drainage radius the difference in drainable volumes between the narrowest and the widest channel scenarios is large, and that for a large well drainage radius the tidal channel width hardly influenced the drainable volume. The sensitivity analysis highlighted the importance of sand-dominated tidal flats in improving lateral connectivity.