Open mazes are primarily designed to measure place learning and memory, using environmental visuospatial cues. However, maze tasks differ along many dimensions, including (1) types of apparatus, which vary from arenas (water maze: WM) to highly structured routes (radial-arm maze: RAM); (2) availability of visuospatial, associative or sensory cues; (3) task requirements which range from spontaneous exploration to complex sequences of choices; and (4) motivation which may involve aversive escape, the opportunity to shelter or to discover novel objects or food at particular locations. Given this diversity, it is likely that mazes tap a variety of processes that contribute to, or affect spatial learning. Hence 'spatial' abilities measured in one procedure may not resemble those engaged in another, posing problems for the interpretation of drug- or lesion-induced deficits. This review compares two types of maze that exemplify key differences in procedure: the RAM and the WM. (1) Visuospatial, associative and sensory factors contributing to place learning in the two mazes are discussed, together with the types of search strategy that they foster, their differing motivation and vulnerability to effects of non-spatial factors, such as stress and training regime. (2) The equivalence of memory processes (acquisition, working and reference memory) assessed in different mazes is considered, and the extent that these may generalize to non-spatial tasks. (3) Differences in application of the two mazes are evaluated. The WM is well-adapted to the study of selective visuospatial factors in place learning and working memory, but less suitable for repeated measures or for assessment of long-term memory deficits. The RAM detects steady-state reference and working-memory deficits, and is suitable for repeated measures, at the expense of precise analysis of the nature of the processes involved.