All products show some variability in test performance. Because of that, manufacturers do their developmental testing after imposing a margin of clearance below the criterion level specified in a mandatory standard. This provides some assurance that future test variation will not drive some individual units beyond the mandatory level. This paper briefly discusses the difficulties in establishing a practical design limit and shows how a standard intended to impose only minimum requirements can end up forcing something more like maximum requirements when statistical variation in test results are taken into account. The matter is illustrated by calculating clearance margins and design limits based on the results of full-scale barrier crash tests of 33 airbag-equipped Mercury cars.