Numerical demand-forecasting methods are based on time series (eg exponential smoothing, transfer- function models) or on causal relationships (eg multivariable regression analysis, trip-generation models). Qualitative methods depend on informed judgement (eg morphological analysis, the Delphi technique). After describing the available techniques, the author argues that the soundest prediction of tourist demand draws on both areas: it contains rigorous quantitative elements and an expert consensus.