Quantitative landslide risk assessment (QRA) presents major challenges to the geoscience community, because it requires the numerical expression of the chance of future landsliding. The difficulty lies in different perceptions about how precise and reliable this numerical expression has to be in order to be of value to the users of a risk assessment. The absence of suitable landslide time series datasets in most regions of the world and the fact that many landslides are essentially one-off events means that it is generally not practical to expect that estimates of landslide probability will be ‘objective’ truths, supported by ‘hard facts’. In most cases the estimation of landslide probability will be based on judgement, making use of the available data, knowledge and experience. It follows, therefore that quantitative landslide QRA cannot be based on ‘objective truths’, relying on the subjective beliefs of the experts involved. In many cases a combination of these two approaches is required. A healthy scepticism is needed when using the results from a landslide risk assessment.