BackgroundRobust information on relative effects of hormonal contraceptives on endogenous androgens is important for understanding beneficial and adverse effects, method choice and development of new methods.MethodsIn this ancillary study at the East London, South Africa site of the ECHO multicentre randomized trial, we compared effects of three contraceptive methods on serum androgen levels among contraceptive users aged 18 to 35 years. Participants were allocated by centrally-managed randomization to open label depot medroxyprogesterone acetate (DMPA-IM), copper intrauterine device (IUD) or levonorgestrel implant.The primary outcome was free testosterone at 6 months.ResultsWe analysed stored baseline and 6-month serum samples in 398/615 participants (DMPA-IM 131/205, IUD 135/205 and implant 132/205). Median testosterone levels at baseline were DMPA-IM 0.82, IUD 0.9 and implant 0.87 nmol/L; at 6 months, DMPA 0.68 (lower than IUD, mean percentage difference 28.35, (p < 0.001), IUD 0.86 (unchanged) and implant 0.66, lower than IUD, mean percentage difference − 22.98, p < 0.001).Median SHBG levels at baseline were DMPA 52.4, IUD 50.5 and implant 55.75 nmol/L; at 6 months, DMPA 40.65, lower than IUD (mean percentage difference 21.19, p = 0.005), IUD 49.1 (unchanged), and implant 23.35 nmol/L, lower than IUD (mean percentage difference − 50.04, p < 0.001 and than DMPA (mean percentage difference − 39.45, p < 0.001).Free testosterone levels at baseline were DMPA 10, IUD 12 and implant 11 pmol/L; at 6 months, DMPA 11, less than IUD (mean percentage difference 13.53, p = 0.047), IUD 12 and implant 14, higher than IUD (mean percentage difference 14.15, p = 0.038) and than DMPA, (mean percentage difference 29.60, p < 0.001).ConclusionsThis is the first randomized trial to show lower SHBG and higher free testosterone with the levonorgestrel implant than with DMPA, and contrasts with reports of increased SHBG with combined oral ethinyl estradiol/levonorgestrel use, and reduced androgens (and impaired sexual function) reported with the etonorgestrel implant. The higher free testosterone with the LNG implant might improve sexual function, mood and bone health as well as increasing side-effects such as acne and hirsutism, and is consistent with the greater sexual activity (with respect to multiple sex partners, new sex partner and unprotected sex) with the implant compared with DMPA documented in the ECHO study.ECHO trial registrationClinicalTrials.gov, number NCT02550067 15/09/2015.Plain English summaryContraception, or family planning, is central to the role of women in societies. It is most important to have accurate information on the relative side-effects of various contraceptive options in order to empower women to make informed choices regarding their preferred method.Hormonal contraceptives contain various forms of the female sex hormones, estrogens and/or progestogens. These hormones have direct effects on the users, as well as modifying the levels of the users’ own circulating sex hormones, both the ‘female’ and the ‘male’ sex hormones (androgens).In this study, consenting participants requesting contraception, were allocated randomly to receive either depot medroxyprogesterone acetate (DMPA-IM) a 3-monthly progestogen injection, the copper intrauterine device (IUD), a non-hormonal contraceptive inserted within the womb, or the levonorgestrel implant, a device placed under the skin which releases a progestogen for 5 years.We measured the participants’ androgen levels after 6 months, and found for the first time that the active form of testosterone (free testosterone) was 29% higher with the implant than with DMPA-IM. The level with the IUD was intermediate, and significantly different from the other two methods.This finding is relevant to the effects experienced by users of these methods, because free testosterone has effects on sexual function, bone health and mood, as well as on conditions such as acne and hair distribution patterns.