Boundary value analysis is the technique of making sure that behavior of system is predictable for the input and output boundary conditions. Reason why boundary conditions are very important for testing is because defects could be introduced at the boundaries very easily. For example, if you were to write code to simulate following condition -

"Input should be greater than equal to 10 and less than 50"

Probably you will write something like

if (input >=10 AND input <50) then

    do some thing

else

do some thing else.

So, according to this input values from 10 to 49 are valid, but if you make mistake in specifying the conditions, following things can happen

• input >10 AND input Input value 10 in invalid now.
• input <=10 AND input Input value 9 is valid now.
• input >=10 AND input Input value 50 is valid now
• input >=10 AND input >50 -----> Input value 49 is invalid now

Because it is very easy to introduce defects at boundaries, boundary values are important. So for the above example, at the minimum we should have following test cases for boundaries

9, 10, 11 and 48, 49, 50

As a general rule, if you have any input field it should be tested with following set of data at the minimum.

lower_boundary - 1, lower_boundary, lower_boundary + 1 and upper_boundary - 1, upper_boundary, upper_boundary + 1