Reliability is how consistently a test measures what it attempts to measure. Why is consistency important? Well, when you measure something with an instrument two times, you want it to come out with the same answer (or close to it) both times.

With the MBTI® instrument, as with other psychological instruments, you want the person to come out the same type both times they take it (this is test-retest reliability, the kind most people care about). Because personality is “slippery” to measure, psychological instruments cannot have the same consistency you would expect from, say, a ruler. But there are generally accepted standards for psychological instruments. . . . It should be understood that the MBTI® instrument meets and exceeds the standards for psychological instruments in terms of its reliability. There is also a kind of reliability that addresses the degree to which someone answers questions consistently on any given scale on the same taking of the MBTI® instrument. This is, not surprisingly, called internal consistency reliability. This is of special interest to people who construct instruments because the more consistency there is, the less “noise” there is in the measurement process. It is of interest to (MBTI®) practitioners because it tells us that there is more “noise” when using the MBTI® instrument with some groups of respondents — and this is important to know.