This paper describes the evaluation of a model for predicting the drift of iceberg ensembles. The model was developed in preparation for providing an iceberg forecasting service off the Canadian east coast north of about 45°N. It was envisaged that 1–5 day forecasts of iceberg ensemble drift will be available. Following a critical examination of all available data, 10 data sets containing up to 404 icebergs in the Grand Banks area off Newfoundland were selected for detailed study. The winds measured in the vicinity of the study area as well as the detailed current system developed by the International Ice Patrol were used as inputs to the model. A discussion on the accuracy and limitations of the input data is presented. Qualitative and quantitative criteria were used to evaluate model performance. Applying these criteria to the results of the computer simulations, it is shown that the model provides good predictions. The degree of predictive success varied from one data set to another. The study demonstrated the validity of the assumption of random positioning for icebergs within a grid block, especially for ensembles with large numbers of icebergs. It was found that an “average” iceberg size can be used to represent all icebergs. The study also showed that in order to achieve improved results it will be necessary to account for the deterioration (complete melting of icebergs), especially during the summer months.