[en] Many human vigilance experiments are summarized and principles are extracted which should be useful in designing and evaluating a nuclear material safeguard system. A human is a poor observer and is not a dependable part of any man-machine system when required to function as an observer. There are a few techniques which improve his performance by providing feedback. A conceptual model is presented which is helpful in design and evaluation of systems. There is some slight experimental support for the model. Finally, some techniques of time study and statistical control charting will be useful as a means of detecting nuclear diversion attempts