Creep is the time dependent deformation of concrete induced by sustained stress. Several different prediction models are currently used throughout the world, however little is known about their ability to predict creep in high strength concretes. Several of these models were designed to be used on normal strength concretes. This text aims to analyse and modify the models where necessary so that they can be applied to high strength concrete. The results are then compared to three sets of measured data. The prediction methods which are compared are the revised AS3600, GL2000, ACI, Eurocode 2 and B3 models. These were chosen because they are the most commonly used prediction models in concrete design.Models which are designed for use on normal strength concrete incorporate both a basic and drying creep component. High strength concretes differ as they are not affected by drying creep and therefore only the basic creep component of the model should be considered (Warner et al. 1998). The basic creep component is isolated and predictions are made based on this modification. It has been found that the removal of drying creep has a positive effect on the predictions.Each of the models are compared to measured laboratory data and the most accurate methods are found to be the B3 and Eurocode 2 models. Eurocode 2 is found to be the most effective as it is easier to apply and the input parameters required are all available at the design stage. B3 is found to be complicated and requires parameters which would be difficult to identify. The revised AS3600 model is found to over predict measured data despite the fact that it was designed to be applicable over both normal and high strength concrete. The ACI model is found to be greatly improved when drying creep is removed for concrete strengths in the order of 90MPa but not for 60MPa, however it requires inputs not available at the design stage. The GL2000 model is found to have a vast improvement when the modification is applied; however the basic creep component is based only on age at loading and therefore results may not be as accurate for different conditions. Eurocode 2 is found to be the most effective model based on accuracy, simplicity and availability of input parameters.