No formal epidemiological research based on systematic analysis has focused on the relationship between herpes zoster and immunodeficiency virus (HIV) infection in Taiwan. Our aim was to explore whether herpes zoster is an early manifestation of undiagnosed human HIV infection in Taiwan. Methods: This was a retrospective cohort study using the database of the Taiwan National Health Insurance Program. A total of 35,892 individuals aged ≤ 84 years with newly diagnosed herpes zoster from 1998 to 2010 were assigned to the herpes zoster group, whereas 143,568 sex-matched and age-matched, randomly selected individuals without herpes zoster served as the non-herpes zoster group. The incidence of HIV diagnosis at the end of 2011 was estimated in both groups. The multivariable Cox proportional hazards regression model was used to estimate the hazard ratio and 95% confidence interval (CI) for risk of HIV diagnosis associated with herpes zoster and other comorbidities including drug dependence and venereal diseases. Results: The overall incidence of HIV diagnosis was 4.19-fold greater in the herpes zoster group than that in the non-herpes zoster group (3.33 per 10,000 person-years vs. 0.80 per 10,000 person-years, 95% CI 4.04–4.35). The multivariable Cox proportional hazards regression analysis revealed that the adjusted hazard ratio of HIV diagnosis was 4.37 (95% CI 3.10–6.15) for individuals with herpes zoster and without comorbidities, as compared with individuals without herpes zoster and without comorbidities. Conclusion: Herpes zoster is associated with HIV diagnosis. Patients who have risk behaviors of HIV infection should receive regular surveillance for undiagnosed HIV infection when they present with herpes zoster.