Background: Fat lipoatrophy improves when a thymidine analog is replaced with abacavir (ABC) or tenofovir DF (TDF). Choice of replavement may be influenced by lipid changes and other safety issues. Objective: To evaluate lipid changes by baseline thymidine analog and replacement agent over 48 weeks. Methods: Fasting lipid data were obtained in a randomized, open-label, 48-week study of AZT or d4T with ABC (300mg BD) or TDF (300mg QD) in adults on HAART with moderate/severe lipoatrophy who are naive to ABC and TDF, and have a current viral load <50cp/ml. Results: 105 adults receiving d4T (n=71) or AZT (n=34) were randomized, 52 to TDF, 53 to ABC. At week 48 there was a significant increase in limb fat in the both groups (p<0.01) but no difference between drug arms. Median baseline lipids for TDF and ABC arms were Total Cholesterol (TC) 5.6 and 5.3, HDL 1.3 and 1.3, LDL 3.3 and 3.0 and triglycerides (TG) 2 and 1.7mmol/l, respectively. All lipid parameters showed a trend to be higher in persons on d4T at baseline. Overall, median lipid parameters improved modestly from baseline with switching to TDF (TC -0.1, HDL -0.01, LDL -0.1, TG -0.17mmol/l) but were unchanged with abacavir. A decline was seen in TC from baseline to week 48 in those on d4T at entry and switched to TDF were from 5.9 to 5.4 mmol/l while no change was seen with switch to ABC (TC 5.4 at both points). In individuals on AZT at entry small rises in median TC were observed; TDF 4.9 to 5.1 and ABC 5.3 to 5.9mmol/l, baseline to week 48, respectively. Conclusions: In lipoatrophic HIV-infected adults, switching from a thymidine analog to ABC or TDF for 48 weeks improves lipoatrophy. Modest benefits on lipids are observed in the TDF group. Lipid benefits are mainly seen in persons switching from d4T to TDF.