ImportanceConcerns that take-home naloxone (THN) training may lead to riskier drug use (as a form of overdose risk compensation) remain a substantial barrier to training implementation. However, there was limited good-quality evidence in a systematic review of the association between THN access and subsequent risk compensation behaviors.ObjectiveTo assess whether THN training is associated with changes in overdose risk behaviors, indexed through injecting frequency, in a cohort of people who inject drugs.Design, Setting, and ParticipantsThis cohort study used prospectively collected self-reported behavioral data before and after THN training of participants in The Melbourne Injecting Drug User Cohort Study (SuperMIX). Annual interviews were conducted in and around Melbourne, Victoria, Australia, from 2008 to 2021. SuperMIX participants were adults who regularly injected heroin or methamphetamine in the 6 months preceding their baseline interview. The current study included only people who inject drugs who reported THN training and had participated in at least 1 interview before THN training.ExposureIn 2017, the SuperMIX baseline or follow-up survey began asking participants if and when they had received THN training. The first THN training date that was recorded was included as the exposure variable. Subsequent participant interviews were excluded from analysis.Main Outcomes and MeasuresInjecting frequency was the primary outcome and was used as an indicator of overdose risk. Secondary outcomes were opioid injecting frequency, benzodiazepine use frequency, and the proportion of the time drugs were used alone. Fixed-effects generalized linear (Poisson) multilevel modeling was used to estimate the association between THN training and the primary and secondary outcomes. Time-varying covariates included housing status, income, time in study, recent opioid overdose, recent drug treatment, and needle and syringe coverage. Findings were expressed as incidence rate ratios (IRRs) with 95% CIs.ResultsThere were 1328 participants (mean [SD] age, 32.4 [9.0] years; 893 men [67.2%]) who completed a baseline interview in the SuperMIX cohort, and 965 participants completed either a baseline or follow-up interview in or after 2017. Of these 965 participants, 390 (40.4%) reported THN training. A total of 189 people who inject drugs had pretraining participant interviews with data on injecting frequency and were included in the final analysis (mean [SD] number of interviews over the study period, 6.2 [2.2]). In fixed-effects regression analyses adjusted for covariates, there was no change in the frequency of injecting (IRR, 0.91; 95% CI, 0.69-1.20; P = .51), opioid injecting (IRR, 0.95; 95% CI, 0.74-1.23; P = .71), benzodiazepine use (IRR, 0.96; 95% CI, 0.69-1.33; P = .80), or the proportion of reported time of using drugs alone (IRR, 1.04; 95% CI, 0.86-1.26; P = .67) before and after THN training.Conclusions and RelevanceThis cohort study of people who inject drugs found no evidence of an increase in injecting frequency, along with other markers of overdose risk, after THN training and supply. The findings suggest that THN training should not be withheld because of concerns about risk compensation and that advocacy for availability and uptake of THN is required to address unprecedented opioid-associated mortality.