Introduction: High proportion of non-testers for HIV has been a main obstacle in estimation of HIV prevalence from household and population based surveys. The objective of this study is to analyze population based surveillance data using a simple model in order to estimate period prevalence of HIV for residents who did not test for HIV by using age specific mortality rates. Methods: This is an open cohort study established since 1994 where baseline survey was conducted in Kisesa, Mwanza northwest of Tanzania. Kaplan-Meier curves were used to compare cumulative probability of deaths for tested and untested individuals and equality of survival curves was tested using Log-rank test. Cox regression analysis was employed to estimate hazards of death for the tested and untested residents. An analytical model was used to estimate period prevalence of HIV for residents who did not test for HIV. Results: Mortality rates for HIV positive residents were >10 times higher than that of HIV negative for all survey years while mortality rates of untested residents was almost three times as high as those who tested negative. Residents who did not test for HIV had significantly higher period prevalence of HIV than those who tested for HIV across all aged groups and sex. Residents who did not test for HIV had 4 times more hazards of death than tested residents (HR=3.8, 95%C.I.=3.3-4.4) adjusted for age and sex. Conclusion: The simple model used for this analysis proved to be an appropriate model to estimate period prevalence of HIV in a demographic surveillance system with high proportion of residents not testing for HIV given that prospective mortality data are available and reliable.