Early identification of patients co-infected with HIV and human T-lymphotropic virus type 1 (HTLV-1) is essential to improve care, as CD4+ T-cell counts have been revealed to be an unreliable laboratory parameter to monitor HIV infection in co-infection. Unfortunately, HTLV-1 testing is not currently available in sub-Saharan Africa. We conducted this study to determine the performance of absolute CD4+ T-cell count estimation in guiding the clinical suspicion of co-infection. A cross-sectional survey was conducted in antiretroviral-naïve HIV (AN-HIV) patients attending an HIV outpatient clinic in Maputo city, Mozambique. Seven hundred and one AN-HIV patients were enrolled in the study. The prevalence of HTLV-1 co-infection was 4.5% (95% confidence interval [CI] 3.0-6.0%). Logistic regression analysis showed that CD4+ T-cell count was an independent predictor of co-infection (P value: 0.000). The performance of absolute CD4+ T-cell counts in predicting co-infection was higher in symptomatic HIV patients when compared with asymptomatic HIV patients. The best performance was achieved with the cut-off of CD4+ count of 500 cells/mm(3), which gave sensitivity, specificity, positive and negative predictive values of 54.2%, 87.2%, 24.0% and 96.2%, respectively. In conclusion, our data provide evidence that the absolute CD4+ T-cell count is of moderate accuracy in guiding the clinical suspicion of co-infection in AN-HIV and its implementation could improve the care provided to a significant number of HIV patients in Mozambique.