We examined the utility of estimating rates of automobile seat belt use with self-report measures. Self-report measures overestimate belt use rates compared to observational surveys of the same population. Laws mandating seat belt use did not substantially affect the degree to which self-reports are upwardly biased. We found self-report measures overestimate observed belt use by 8.9 to 19.4 percentage points or by a factor of 1.2 to 2. Our best estimate is that self-reported seat belt use rates be discounted by 12 percentage points to estimate actual belt use rates.