In 1983 and 1984, surveys were conducted in four Chinese population samples, urban and rural for both Beijing and Guangzhou, as part of PRC-USA collaborative research in cardiovascular and cardiopulmonary epidemiology. Serum total cholesterol (TC), HDL-cholesterol (HDL-C), and triglycerides (TG) were measured in 4280 men and 4695 women aged 35-54 years, in laboratories standardized by the US Centers for Disease Control. Age-adjusted mean serum TC was higher in urban than rural samples and generally higher in Beijing than Guangzhou, ranging from 155 mg/dL for Guangzhou rural women to 187 mg/dl for Guangzhou urban women. Group mean values of HDL-C varied from 48 to 59 mg/dl, higher in Beijing than Guangzhou, and higher in women than men. TC/HDL-C ratio ranged from 3.05 to 3.82. Serum TG values were higher for Beijing than Guangzhou; the lowest group mean values of 78 mg/dl and 75 mg/dl were in rural Guangzhou men and women. Mean body mass index (BMI) was uniformly low, ranging from 20 kg/m2 for rural Guangzhou to 24 kg/m2 for urban Beijing. Multiple regression analyses showed that BMI was positively and independently related to serum TC, LDL-C, TG, and TC/HDL-C, and inversely related to HDL-C. Smoking was positively related in both sexes to TG and TC/HDL-C, and inversely related to HDL-C. Smoking was also positively related to TC and LDL-C in men. In men, alcohol was positively related to TC and HDL-C, and was inversely related to TG and TC/HDL-C. Heavy manual work was inversely related to TC, LDL-C, and TC/HDL-C in men, but not related to lipids in women. Thus, for these Chinese population samples, despite their lower serum TC and BMI, the correlates of serum lipids are similar to those in western populations. These variables accounted for only part of the observed urban-rural and north-south differences in serum lipids among these Chinese population samples. The significance of the relatively low serum TC and TG and high HDL-C in relation to low cardiovascular disease in Chinese populations is the object of further investigation in follow-up studies.