The author examined the relationship between C-reactive protein and BMI and diabetes status among 16,573 participants aged > or = 20 years of the Third National Health and Nutrition Examination Survey (1988-1994). The study had a cross-sectional design. Geometric mean concentrations of C-reactive protein were lowest among individuals with a BMI < 18.5 kg/m2 and increased with increasing BMI categories. Restricting the analysis to participants without various medical conditions did not change the relation. After adjusting for age, sex, race or ethnicity, and education, using logistic regression analysis, odds ratios for an elevated C-reactive protein concentration (> or = 85th percentile of the sex-specific C-reactive protein concentration distribution) among participants with a BMI of 25 to < 30, 30 to < 35, 35 to < 40, and > or = 40 kg/m2 were 1.51 (95% CI 1.23-1.86), 3.19 (2.60-3.91), 6.11 (4.67-7.98), and 9.30 (6.43-13.46), respectively, compared with participants with a BMI < 25 kg/m2. C-reactive protein concentrations were lowest among those individuals without diabetes or with impaired fasting glucose and highest among those with newly or previously diagnosed diabetes. Compared with participants with a normal fasting glucose, participants with impaired fasting glucose, newly diagnosed diabetes, and previously diagnosed diabetes had 0.99 (0.72-1.37), 1.84 (1.25-2.71), and 1.59 (1.25-2.01) odds of having an elevated C-reactive protein concentration after adjustment for age, sex, race or ethnicity, education, and BMI. These results confirm cross-sectional findings from previous studies that show elevated C-reactive protein concentrations among individuals who are obese or have diabetes. The implications of these findings, however, remain unclear.