Thirty human immunodeficiency virus (HIV)-infected men were randomized to a high dose of nandrolone decanoate weekly (group 1) or nandrolone plus resistance training (group 2) for 12 wk. For the two groups, nandrolone had no significant effects on total cholesterol, LDL cholesterol, LDL phenotype, or fasting triglycerides, although triglycerides decreased by 66 +/- 124 mg/dl for the entire population (P = 0.01). Group 2 subjects had a favorable increase of 5.2 +/- 7.7A in LDL particle size (P = 0.03), whereas there was no change in group 1. Lipoprotein(a) decreased by 7.3 +/- 6.8 mg/dl for group 1 (P = 0.002) and by 6.9 +/- 8.1 for group 2 (P = 0.013). However, HDL cholesterol decreased by 8.7 +/- 7.4 mg/dl for group 1 (P < 0.001) and by 10.6 +/- 5.9 for group 2 (P < 0.001). Percentages of HDL(2b) (9.7-12 nm) and HDL(2a) (8.8-9.7 nm) subfractions decreased similarly for the two groups, whereas HDL(3a) (8.2-8.8 nm) and HDL(3b) (7.8-8.2 nm) increased in the groups during study therapy (P < or = 0.02 for all comparisons). There was no evidence of a decreased insulin sensitivity in either group, whereas fasting glucose, fasting insulin, and homeostasis model assessment improved in group 2 (P < 0.05). These metabolic effects were favorable (other than for HDL), but changes were generally transient (except for HDL in group 2), with measurements returning to baseline 2 mo after the interventions were completed.