Adiponectin is an adipocyte-derived protein with anti-diabetic, anti-atherogenic and anti-inflammatory action, but there are few studies on its association with cardiovascular and metabolic risk factors in different ethnic groups in Australia. This cross-sectional study evaluated ethnic differences in adiponectin levels and its association with age, gender, body composition and diet in 89 adult Australians of European (n = 28), Indian (n = 28) and Iranian (n = 33) ancestries. Different measures of adiposity were assessed using the method of whole body dual energy X-ray absorptiometry (DEXA). Total adiponectin levels determined in Indians and Iranians were significantly lower than those in Europeans (p values < 0.001). There was no significant difference between the adiponectin levels in Indians and Iranians (p value > 0.05). There was no substantial change in the results after adjustment for potential confounders. Circulating levels of adiponectin was associated with age, truncal fat percentage, dietary glycemic index, glycemic load and carbohydrate intake, by correlation analysis (p values < 0.05). Using multiple linear regression analysis, a model including truncal fat percentage (p < 0.001), ethnicity (p = 0.001), age (p = 0.001) and dietary glycemic index (p = 0.04) could predict 50% of the variance in adiponectin levels (R2 = 0.504). Among different variables assessed, truncal fat percentage (in Indian and Iranian groups) and glycemic index (in European group) were the strongest predictors of serum adiponectin when data were analysed for three ethnic groups, separately. In conclusion, individuals with Iranian or Indian ancestries may have lower adiponectin levels compared to Europeans. Ethnicity was found as an independent factor affecting adiponectin levels. Our results also highlighted age, truncal adiposity and dietary glycemic index as other determinants of serum adiponectin, however the extent to which these factors influence adiponectin concentrations may vary across ethnicities.