Many authors have used a Kalman Filter (KF) or a bank of several KF's as the main component of a fault detection algorithm (see, e.g., [5, 9]). Usually, the residual (or error) from the KF is evaluated against a predetermined threshold and crossing of the threshold level triggers a fault flag. The exact nature of the residual evaluation varies from analysis of the raw signal, to application of relatively complex statistical tests [2, 9, 11]. However, it is not clear from the literature which of the many methods available offer the best results. The paper examines the application of several statistical tests to residuals of a KF implemented as part of a fault detection scheme on an aircraft fuel system simulator test-rig. The experimental results will be evaluated and discussed and recommendations will be made on which methods offer the greatest utility for rapid detection of a leak fault applied to a tank containing fluid on the test-rig. The statistical methods evaluated are: mean deviation, mean absolute deviation, mean square error, root mean square error, sum of square error, weighted sum of square error, paired-t test, r-square and chi-square mean.