To determine whether ultra-low doses of estradiol (E(2)) affect the serum lipid profile in elderly women, we analyzed changes in serum lipids and lipoproteins in 70 healthy women, 60 yr and older, randomly assigned to parenteral E(2) (7.5 microg per 24 h) delivered by a vaginal ring (Estring; Pharmacia-Upjohn, Malmö, Sweden) or no treatment for 12 months. Baseline serum estrone sulfate (E1S), but not E(2) or serum FSH, was negatively associated with serum total cholesterol (P = 0.026), low-density lipoprotein (LDL) cholesterol (P = 0.053), and apolipoprotein B levels (P = 0.023). Compared with no treatment, Estring treatment yielded nonsignificant increases within the normal postmenopausal range in serum E1S (+16%) and E(2) (+13%), but significantly reduced serum LDL cholesterol by 7.6% (-0.32 mmol/L; 95% confidence interval, -0.58, -0.07; P = 0.014) and LDL to high-density lipoprotein (HDL) ratio by 7.3% (-0.19 mmol/L; 95% confidence interval, -0.44, -0.06; P = 0.030). In Estring users values were significantly reduced in total cholesterol (by 4%), LDL cholesterol (by 7%), LDL to HDL ratio (by 7%), and apolipoprotein B (by 4%), and significantly increased in serum HDL triglyceride (by 25%) but not triglycerides. No significant changes were found in the untreated group. There was a significant interaction between age and both baseline serum E(2)/sex hormone-binding globulin (P = 0.006) and sex hormone-binding globulin (P = 0.009) and a marginal interaction between age and E1S (P = 0.083) with regard to effects on changes in LDL cholesterol levels during Estring treatment. We conclude that ultra-low doses of E(2), which previously were considered to have only local effects, may improve serum lipid profile in elderly women with a pattern and magnitude similar to that reported after conventional estrogen doses or first-generation lipid-lowering agents. The reduction in LDL cholesterol tended to be greater with a combination of high age and low baseline levels of biologically active estrogens.