Some of the key summary points and conclusions raised by this work are listed below: (1) A modified one zone model has been proposed to predict the mean temperature of steel members subjected to post-flashover fires. In the model, a quantity which considers the heat sink effect of steel members in the fire compartment is added to the heat balance equation for the traditional one zone compartment fire model. Based on the modified one zone model, numerical or analytical approaches can be developed to give more rational prediction of steel and gas temperatures in post-flashover fire conditions. (2) Current formulae provided by fire codes in different countries for calculating the steel temperature of insulated steel members in fire conditions are based on the standard fire model, which might give unacceptable results for calculation in natural fires. Besides, when using those formulae, iterative computations should be always processed, which is not convenient for daily design works and is not efficient for reliability analysis which usually includes hundreds of thousands of simulation loops for a single case. (3) Asimple approach has been developed for calculating maximum steel temperature of insulated steel members in natural fires. The approach adopts time equivalent to relate natural fires with the standard fire, and use a simple quadratic equation for calculating the maximum steel temperatures. By comparing with numerical results and test data, the proposed approach can give satisfactory prediction of maximum steel temperatures in the range from 300 to 600 °C. The professor factor of the approach has been characterized by test data, which has a mean of 0.955 and a COV of 0.014, and can be best described by lognormal distribution. The approach only needs hand calculations which is easy and convenient for practical usage, and the approach is given in a closed form which is efficient for reliability analysis. (4) Intumescent coatings will react at high temperatures and the thermal properties of intumescent coatings can not be measured directly by the current standard test methods which are originally designed for the traditional inert fireproofing materials. A simple procedure has been proposed to assess the fire resistance of intumescent coatings by using the concept of equivalent constant thermal resistance. By using the equivalent constant thermal resistance derived at a critical steel temperature of 550 °C, the calculated steel temperatures agree well with the test data in the range of the concerned limiting temperatures from 400 to 600 °C. (5) In practice, when specifying coating fire protection for steel structure, it assumes that the coating is correctly applied and its performance meets the fire protection needs without degradation over time. However, since the organic components of intumescent coating, it should be expected that the fire protection function of intumescent coating over time would not be as reliable as when freshly applied. The values of constant thermal resistance of two thickness of intumescent coating exposed to aging condition reported in literature are derived. It shows that aging has effect of reducing the constant thermal resistance of intumescent coatings. (6) Aprobabilistic approach has been provided to determine the service life of intumescent coatings for steel columns. The approach compares the failure probability of the protected steel columns with the target probability of the structural fire design. The probability of fire occurrence and the probability of flashover are considered and determined from codes. An example is given to determine the service life of ICs for protecting steel columns in an office building. The service life for the intumescent coatings in the example is 20 years. The approach is also applicable for designing the traditional inert fire proofing materials.