Large timber buildings, formed from both light and heavy timber construction, are becoming increasingly common in Europe. Many multiple-occupancy timber buildings, such as apartment blocks, are now constructed to greater heights and in densely populated urban locations. The fire-resistance performance of such timber buildings is generally related to the standard fire test. Alternatively, EN 1995-1-2 may be used to demonstrate fire resistance by means of calculation or numerical modelling. The latter is currently limited to standard fire exposure. In addition, modelling approaches are often avoided as many numerical codes do not normally offer the capability to model timber exposed to fire. The most obvious barrier is incorporating the different tensile and compressive strength/stiffness degradation with increasing temperature. Unlike many other structural materials, it is not possible to define a single relationship between timber Modulus of Elasticity (MoE) and temperature. When timber design is advanced to a ‘performance-based’ level further complexities will arise. For example, the definition of structure temperatures for non-standard fires is a difficult task, and assessment of strength/stiffness degradation on the basis of temperature alone is not sufficient due to char formation. As a result, when cooling is considered, material properties based upon stress state, temperature and temperature history are needed.To address the above limitations, a number of developments, which can be used with general FEA software, such as DIANA, to design timber structures for fire, are presented. The developments are incorporated via user-supplied subroutines written in FORTRAN code. The routines include code for determining MoE and strength based upon stress state, temperature and temperature history. They are implemented as part of a total strain-based constitutive model. The implementation of the routines is demonstrated using a simple continuous beam. The example is also used to demonstrate how compartmentation provisions and aspects of whole building behaviour can be used to better design large-section timber buildings. Comparisons are made with simple empirical approaches presented in EN 1995-1-2. Extensions to ‘performance-based design’ using parametric fires are also discussed.