A simulator experiment was conducted to determine the potential benefits of path prediction on the navigation performance with channel-bound vessels. Six pilots had to sail an approach channel by a deep draught vessel under critical conditions. For navigation, a basic ARPA radar system was used, supplemented by three types of path predictors. One predictor (a) was based on an accurate fast-time hydrodynamic model of the vessel and showed the exact path of the vessel. Both other path predictors were less accurate, relatively simple extrapolators. Predictor (b) was based on a non-linear speed and rate of turn extrapolator, predictor (c) on a linear speed and course extrapolator, showing the ground velocity vector. The navigation performance was determined in terms of deviation from the planned route. The results indicated that the best navigation performance was obtained by including the vessel's rate of turn in the predictive model, as was the case with path predictor (a) and (b). The navigation accuracy increased a factor two compared to linear extrapolation (predictor type (c)). Implications of the use of path prediction for channel-navigation are discussed.