Building energy assessment methods, used in a regulatory context, impose a calculation procedure under restricted and predefined conditions to check preset energy performance levels. Standardized boundary conditions and input data are implemented allowing for objective evaluation of various building designs. Focusing on school buildings in particular, these specific boundary conditions are however often inaccurate or even unavailable. Therefore, throughout this paper, typical school characteristics and their uncertainty are studied. The impact on the energy demand calculations is then demonstrated in an uncertainty analysis using the Monte Carlo analysis method combined with the Latin Hypercube sampling technique. A sensitivity analysis, using the elementary effect method of Morris, reveals the set point temperature, load and users' profiles of lighting and equipment in class rooms as the most dominant input parameters on which additional surveys and questioning are focused to determine possible inaccuracies and to define, where necessary, new, more realistic boundary conditions. As a final result, a list of representative boundary conditions for Flemish schools is determined which can be used directly or can be converted into averaged input data for simplified calculation methods. (C) 2014 Elsevier B.V. All rights reserved.