In evaluation, the characteristic of a testing instrument that yields consistent results each time it is used.
In research, the reproducibility of an experimental result, how closely a second go-around would yield the same answer whether or not correct.
In relation to tests, the degree to which a particular test gives the same scores or readings consistently. For an intelligence test, reliability would concern whether or not a student got approximately the same score on a different form of the test given at a different time (this is test-retest reliability) or whether conditions in the testing situation, variability in the tests, or change in the person giving the test (interobserver reliability) would produce widely varying results. For a medical test, reliability primarily concerns how often a test gives accurate results, as opposed to false positives (suggesting a problem where none exists) or false negatives (showing no problem when one does exist).
The degree of accuracy of results over a period of time, a number of trials, or among different observers or investigators. Also, the probability that a system will perform its function properly for a given period of time.