The need for assessment of regional vulnerability of groundwater to pollution from surface activities is highlighted. It is generally considered that efficient broad-scale assessments should be based on a small number of readily available spatial coverages which delineate hydrogeological settings. A variety of approaches reported in the literature is reviewed. These include empirical (e.g. the DRASTIC system), deterministic, probabilistic, and stochastic techniques. Empirical approaches have been criticised for being subjective, whilst physically based deterministic techniques are more appropriate for sub-regional predictive (quantitative) assessment and are tied to specific perceived processes. Probabilistic techniques are data intensive and involve groundwater numerical modelling as they are thus inappropriate for broad, regional vulnerability assessments based on readily available spatial variables. It is concluded that stochastic techniques offer more promise than other techniques involving weights-of-evidence modelling; new work in this area is described.