We assess the reliability of Lagrangian simulations based on HF-radar velocity fields upon drifter trajectories in the Gulf of Trieste, Italy. In particular, we carry out different types of simulations, such as single-particle tracking, lagrangian barrier identification and absolute and relative statistics. The analyses aim at evaluating how the former simulations compare with CODE drifter trajectories and the latter are influenced by initial conditions. Drifter simulations do present increasing errors when the quality of the HF-radar velocity fields deteriorates due to recording issues. However, the joint use of lagrangian barriers and single-particle tracking simulations could be an asset in increasing the quality of our predictions. Besides, initial conditions can influence the outcome of simulations since absolute and relative dispersion can vary up to one order of magnitude depending on the initial conditions set at the beginning of the calculations, i.e. their starting time. The dispersion regimes assessed through finite-size Lyapunov exponents do show such a dependence and some caution should be posed when employing such measures.