The key motivation for Hybrid Electric Vehicles (HEV) is fuel consumption optimisation, but the impact of their control strategies on emissions can be very strong, both positive and negative. Indeed, the Energy Management Problem (EMP) in HEVs consists in optimally splitting power demands among the available energy sources (i.e. batteries and fuel) such that consumption and/or emissions are minimised. While fuel consumption minimisation has been widely studied, simple map-based emissions simulations which assume a quasi-static engine behaviour have been mostly used for emission assessment. This tends to underestimate emissions, and can be particularly wrong in the case of switching strategies, as for most automatic engine start&stop systems, but also for optimal solutions arising from the fuel efficiency minimisation problem solved by minimisation of the total energy costs. This paper proposes an Engine-in-the-Loop-based HEV model for optimal control strategies evaluation which does not require emissions and fuel consumption models for the evaluation. For this purpose, a dynamic forwards (i.e. causal) non-linear power train is modelled, calibrated wit dynamic measurements from corresponding production vehicle. Then, a real engine is placed within the simulation, allowing to measure pollutants and fuel consumption directly from test bench sensors. Thus, a real-time capable simplified linear HEV control strategy is evaluated with real driving cycles as well as offline methods like dynamic programming for benchmarking purposes. The experimental effect of optimally solving the fuel minimisation EMP on pollutant emissions is also considered and evaluated for different scenarios.