Hepatitis C virus (HCV) causes substantial morbidity and mortality, particularly among people who inject drugs (PWID). While elimination of HCV as a public health problem may be possible through treatment-as-prevention, reinfection can attenuate the impact of treatment scale-up. There is a need to better understand the distribution and temporal trends in HCV infection risk, including among HCV-seropositive individuals who will be eligible for treatment and at risk for subsequent reinfection. In this analysis of 840 seronegative and seropositive PWID in Baltimore, MD USA, we used random forest methods to develop a composite risk score of HCV infection from sociodemographic and behavioural risk factors. We characterised the individual heterogeneity and temporal trajectories in this composite risk score using latent class methods and compared that index with a simpler, conventional measure, injection drug use frequency. We found that 15% of the population remained at high risk of HCV infection and reinfection by the composite metric for at least 10 years from study enrolment, while others experienced transient periods of moderate and low risk. Membership in this high-risk group was strongly associated with higher rates of HCV seroconversion and post-treatment viraemia, as a proxy of reinfection risk. Injection frequency alone was a poor measure of risk, evidenced by the weak associations between injection frequency classes and HCV-associated outcomes. Together, our results indicate HCV infection risk is not equally distributed among PWID nor well captured by injection frequency alone. HCV elimination programmes should consider targeted, multifaceted interventions among high-risk individuals to reduce reinfection.