This study estimates adherence and identifies predictors of good adherence among 1305 Tanzanian women participating in a randomised, double-blind, placebo-controlled trial of HSV suppressive therapy to reduce HIV incidence or genital HIV shedding. Women were randomised to acyclovir 400mg BD or placebo and followed every three months for 12-30 months. Adherence was assessed by tablet counts. Random urine samples, collected between 6 and 24 months, were tested for acyclovir. At 12, 24 and 30 month visits, 56%, 52% and 54% of women on treatment had adherence >or=90%, respectively. Factors independently associated with good adherence (taking >or=90% of tablets in the preceding 3-months) included older age, understanding trial concepts at enrolment, living >2 years in the screening site, receiving an unannounced tablet check visit, using oral contraception at screening, living in the same site and house as the previous visit, accessing VCT during the trial, recent malaria and not having a positive pregnancy test. Overall, 55% of urine samples from women randomised to acyclovir had detectable acyclovir. Additional, tailored adherence strategies may be needed for younger, more mobile women and those who have not used oral contraception, which may sensitise them to daily tablet-taking. Use of biomarkers may alert investigators to adherence problems.