The assessment of the likelihood of failure of a structure is a difficult matter which has for too long been neglected by the profession at large. It is surprising, for instance, that statistics relating to structural damage produced by various causes such as fire, explosions, variations in loading and strength effects, are not more readily available. It is encouraging that work is now proceeding, for example, at the Fire Research Station on the likelihood of fire damage. The probability of occurrence of a particularange of levels of severity of damage as shown by a probability density function can be obtained by looking at the past history of the total population of structure or class of structures. However, the problem of predicting the likelihood of an accident resulting in total or partial collapse of a structure is more difficult. Sir Alfred Pugsley has again produced an interesting and thought-provoking paper. As he has pointed out, resort has to be made to a subjective assessment. This suggestion at first sight may seem a desertion of objective scientific reasoning but of course engineering is not a science and relies heavily on personal experience and subjective assessment. It is obviously important that the type of assessments suggested by Sir Alfred are made by very experienced engineers and it is a direct use of their experience channelled to a new purpose.