Recent UK-based studies have shown a performance gap between the energy performance of buildings calculated using tabulated thermophysical properties of solid walls and that estimated from in-situ measurements. Solid-walled buildings have been targeted by UK Government policies and incentive schemes to meet climate change mitigation targets and improve the efficiency of the building stock, as they are less efficient and more expensive to treat than cavity walls. Since it is common practice to estimate energy use and potential savings for buildings retrofit assuming standard values from the literature, the performance gap may have serious implications on the decision-making and the cost-effectiveness of energy-saving interventions. The aim of this paper is to compare and contrast the results obtained from three different methods for estimating normalised dwelling energy demand: a) the UK energy performance certificate (EPC) method, which uses the standard assessment procedure (SAP) with tabulated inputs (the business as usual case); b) the SAP calculated using empirical air change rates from pressure tests and U-values estimated analysing monitored data with a Bayesian-based dynamic method developed by the authors; c) a normalised annual consumption (NAC) method based on empirical energy consumption data from smart meter and weather data. The analysis is performed on a sample of dwellings from the Energy Saving Trust “Solid Wall Field Trials” dataset. Results show that EPC estimates are systematically higher (between 7.5% and 22.0%) than SAP. Conversely, the NAC displayed a large range of relative differences (between -77% and +99%) compared to the EPC. This raises questions about the relative merits and purpose of the EPC and SAP bottom up methods compared to the smart-meter data-driven NAC method. Further research is suggested using SAP 2009 to isolate the thermal component of energy demand and compare it directly with the NAC component.