For a statistical sample a pair of values that delimit the interval for which there is a certain probability that the true value of some deduced variable lies between those values
A sample of 1000 showed that the correlation between weight and height was 0.0167, with 95% confidence limits of 0.0142 and 0.0183.