Fourteen experiments were conducted at the University of Canterbury using a 1.0m x 1.0m x 1.5m compartment with liquid pool fires. They were conducted to experimentally study, in reduced scale, the conditions that exist in under-ventilated compartments, with a focus on smoke explosions. Ventilation into the compartment was controlled to force the fire to burn to extinction, with the temperatures and fuel mass loss rates being recorded. In the process of determining these conditions, the behavioural properties of various building materials and their limitations showed testament to the difficulties that arise when attempting to control a natural energy. Trying to build a compartment to contain fires of temperatures up to 1100°C, was in itself a testing process. The initial fire resistant building materials used, were found to have adverse effects when repeatedly exposed to fires. Although a smoke explosion was not produced, steady state mass loss values for various ventilation openings were found. These were then used to produce a very good ventilation dependent mass loss trend line. The point of extinction of these fires was found, which will be of use in further research. The fires conducted in the experiments were supplied with minimal oxygen, and all produced excessive quantities of unburnt fuel. This was due to the generation of pyrolyzates as a result of the radiative feedback off the compartment walls, diffusion flame and gas layer. The fire initially burns in the pan, until all oxygen within the compartment has been depleted. At this point the combustion transfers from inside the pan, to burn in the localised vicinity around the pan where oxygen is able to penetrate. The combustion then progresses through a transition stage, in which the flame front moves from the rear of the compartment, to the front. This involves a period where the flames oscillates backwards and forwards from the vent. Steady state burning occurs continuously within the vent. Once steady state burning in the vent is achieved, no degree of ventilation reduction, except full sealed closure, would result in extinction of the fire. A fire occurring at this stage is fully fuelled from the volatilisation of the liquid pool at the rear of the compartment. The mass loss of fuel due to this type of burning is found to be well in excess of that available to ventilation controlled stoichiometric burning. The neutral plane layer was found to be located at approximately the mid-height of the opening and little effect is seen on the location of the neutral plane layer at these low ventilation limits.