Objectives A detailed understanding of intentions and practices related to partner notification (PN) following STI diagnosis can improve control strategies. We assessed participant-level and partner-level factors guiding notification behaviour among men who have sex with men and/or with transgender women (MSM-TW) in Lima, Peru, including discordances between anticipated and actual notification.

Methods Men newly diagnosed with gonorrhoea, chlamydia and/or syphilis between 2012 and 2014 reported recent partners’ characteristics, anticipated PN practices, and actual PN outcomes following diagnosis. Generalised estimating equation Poisson regression analyses assessed factors guiding PN outcomes.

Results Participants (n=150) predominantly identified as homosexual (70%) and moderno (versatile sexual role, 55%); 55% of partners (n=402) were casual. Among all sexual partners, 35% were notified of the STI diagnosis, though only 51% of predicted PN occurred and 26% of actual notifications were unanticipated. 47% of participants notified no partners, while 24% notified all partners. PN was more common with stable versus casual (adjusted prevalence ratio (aPR), 95% CI: 0.53, 0.39 to 0.73) or commercial (aPR, 95% CI: 0.38, 0.12 to 1.21) partners, and among participants who perceived PN as normative among their peers (aPR, 95% CI: 1.96, 1.37 to 2.82). A trend towards greater notification following condom-protected intercourse was observed (aPR, 95% CI: 1.33, 0.98 to 1.81). PN frequency did not differ by type of STI diagnosed.

Anticipated notification predicted actual notification (aPR, 95% CI: 1.67, 1.19 to 2.33) only imperfectly: 81 (54%) participants’ PN practices did not match their anticipated behaviour. Successful notification despite anticipated silence (40 participants, 63 partners) was associated with stable partnerships and a normative perception of PN. Non-notification despite intention (43 participants, 73 partners) frequently occurred among participants reporting exclusively oral sex with the partner or with partners identified as activo (insertive role).

Conclusions Anticipated notification imperfectly reflects actual PN behaviour. Future interventions to improve PN among MSM-TW in Peru need to acknowledge partnership contexts.