In the last few years, significant progress has been made in the understanding of H-mode plasmas (e.g. ion temperature profile stiffness, pedestal physics, etc). Based on this improved understanding, a set of rules (models) comprising a physics picture of the H-mode has been implemented in the ASTRA code in order to improve the understanding of experimental observations and ultimately to provide a predictive capability for ITER complementary to the scaling relations. The model has been verified for consistency with experimental observations in ASDEX-UG and JET plasmas. Numerical coefficients for the transport, required because of simplifications or missing quantitative information, are determined for one plasma (e.g. from JET) and then held constant for all others (JET, ASDEX-UP, ITER). After benchmarking the model to experimental results, it was also applied to ITER. It predicts that Q = 10 can be achieved in ITER but only with at least a 50% deep fuelling contribution (inside the H-mode pedestal). However, in existing machines as well as in our model runs for existing machines, gas puffing is sufficient to achieve the observed density pedestal and line average densities. A second important result from the predictive runs for ITER is that electron energy transport in the plasma core, the neoclassical transport in the pedestal and the CX losses at the plasma edge are important constraints for a better performance. Thus future theoretical and experimental work should concentrate on these areas in order to improve our predictions.