There is a basic distinction between declarative memories, which can be stated verbally, and non-declarative memory, such as how to ride a bicycle, which cannot be expressed in words. With age it is the performance of declarative memory, particularly episodic memory that requires recall of events placed in time, that declines. As memory is not a unitary phenomenon, it should be ideally monitored using a range of tests that reflect theoretical conceptions of the topic. If circumstances demand the use of a single test then a measure of episodic memory is suggested. When it proves only possible to use a rating scale it should be ensured that memory is distinguished from other aspects of cognition and that different types of memory are not confused. The tests used, and the form in which they are used, need to be chosen to be of appropriate difficulty for the sample studied. A major conclusion is that the selection of the measure of memory used in the study of a dietary intervention should never be routine. It is inevitable that the form of the test used will need to be chosen carefully for the population being studied.