There is a lack of consensus about whether researchers and practitioners should use a deterministic single value or probabilistic distribution of values for each input when modelling the life cycle of a building. This study produces a direct comparison of the two approaches by modelling the nonoperational life cycle energy of three buildings deterministically and probabilistically to explore whether the two approaches produce different conclusions. A detailed method describes the model - its structure, formulae and the best-case, typical and worst-case inputs - with supporting references. This detail provides a thorough explanation of why probabilistic modelling suggests that non-operational energy could be 28-44% lower or 48-283% higher than the original values and why a significant shift in the distribution of non-operational life cycle energy is possible. When used deterministically, the model suggests non-operational energy use is greatest during the product phase. However, when used probabilistically, the model highlights the risk that short component lives and long distance transport by road can significantly increase non-operational energy during the use, construction and end of life phases. The study discusses how future modelling should address a number of uncertainties so that it is more useful for researchers and practitioners.