In this study we draw attention to the inherent variability in the results of trial trenching, when taking into account the countless variations in orientation and positioning of trenches. Grids of trial trenches were simulated time and again on the excavation plans of 16 archaeological sites from Flanders, Belgium. Orientation and positioning of the grid layout was shifted randomly, whilst the area coverage varied from 2.5% to 80%. The intersection rates of the archaeological features allow to gain more insight in trends and variability that are inherent to the chosen design of trial trenches. It is assessed how robust a chosen grid layout performs on (multi-period) archaeological sites and how variable these results might be. The most effective layout appears to be a grid with short, parallel and discontinuous trenches or a standard grid, closely followed by 2m wide continuous trenches. Implementing 4m wide trenches reduces the effectiveness of the latter method substantially. When the area coverage of the trenches is below 10%, the results of the archaeological prospection become unreliable and can potentially lead to a substantial over-or underestimation of the actual feature density on the site.