Population monitoring of obesity is most commonly conducted using body mass index (BMI). We test the hypothesis that because of increases in waist circumference (WC) independent of increases in weight, BMI alone detects an increasingly smaller proportion of the population with obesity. METHODS: Australian adults with measured height, weight, and WC were selected from three nationally representative cross-sectional surveys (1989, 1999-2000, 2011-12; n=8313, 5903 & 3904). Participants were defined as having obesity using classifications for an obese BMI (≥30kg·m-2) and substantially-increased-risk WC (≥88cm [women], ≥102cm [men]). Age-standardised prevalence of obesity according to BMI and/or WC, and the proportion of these detected by BMI and by WC were compared across surveys. FINDINGS: Between 1989 and 2011-12, weight and WC increased by 5.4kg and 10.7cm (women), and by 7.0kg and 7.3cm (men). For women and men, 63% and 38% of increases in WC were independent of increases in weight. Over this period, the prevalence of obesity according to BMI and/or WC increased by 25.3 percentage-points for women (18.9% to 44.3%) and 21.1 percentage-points for men (17.1% to 38.2%). The proportion of these detected by BMI decreased for women by 20 percentage-points (77% to 57%) with no change for men. The proportion of these detected by WC increased for women and men by 10 percentage-points (87% to 97%) and 6 percentage-points (85% to 91%) respectively. CONCLUSION: BMI alone is detecting a decreasing proportion of those considered obese by BMI and/or WC. Renewed discussion regarding how we monitor obesity at the population level is required.