Trace Tritium Experiments (TTE) on JET were analysed using Monte Carlo modelling of the neutron emission resulting from the Neutral Beam Injection (NBI) of short (~300ms) Tritium (T) beam blips into reversed shear, hybrid ELMy H-mode and L-mode deuterium plasmas for a wide range of plasma parameters. The calculated neutron fluxes from Deuterium-Tritium (DT) reactions could only be made consistent with all plasmas by applying an artificial reduction of the T beam power in the modelling of between 20 and 40%. A similar discrepancy has previously been observed in both JET and TFTR, although no mechanism has yet been found that could explain such a difference in the measured T beam power. Applying this correction in the T beam power, good agreement between calculated and measured DT neutron emission profiles was obtained in low to moderate line averaged density (ne < 4×1019 m- 3) ELMy H-Mode plasmas assuming that the fast beam ions experience no, or relatively small, anomalous diffusion (Dan<<0.5m /s). However, the modelled neutron profiles do not agree with measurements in higher density plasmas using the same assumption and the disagreement between the measured and calculated shape of the neutron profile increases with plasma density. In this paper it is demonstrated that large anomalous losses of fast ions have to be assumed in the simulations to improve agreement between experimental and simulated neutron profiles, characterized by the goodness of fit. Various types of fast ion losses are modelled to explain aspects of the data, though further investigation will be required in order to gain a more detailed understanding of the nature of those anomalous losses.