There are several ways to measure fatness and obesity, each with its own strengths and weaknesses. The primary measure for tracking the prevalence of obesity has historically been body mass index (BMI). This paper compares long-run trends in the prevalence of obesity when obesity is defined using skinfold thickness instead of BMI, using data from the full series of U.S. National Health Examination Surveys. The results indicate that when one uses skinfold thickness rather than BMI to define obesity, the rise in the prevalence of obesity is detectable 10-20 years earlier. This underscores the importance of examining multiple measures of fatness when monitoring or otherwise studying obesity.