The use of population-based survey data together with sound statistical methods can enhance better estimation of HIV risk factors and explain variations across subgroups of the population. The distribution and determinants of HIV infection in populations must be taken into consideration. We analysed data from the HIV Prevalence and Behaviour Survey in Mozambique aiming to find risk factors associated with HIV infection among Mozambican women. The paper provides a complex survey logistic regression model to explain the variation in HIV seropositivity using demographic, socio-economic and behavioural factors. Results show that women aged 25–29 years, living in female-headed households, living in richer households and those widowed, divorced or not living with a partner have higher odds of being HIV-positive. Findings from our study provide a unique and integrated perspective on risk factors for being HIV-positive among Mozambican women and could support the implementation of programmes aiming to reduce HIV infection in Mozambique.