This study proposes a modified Preston-Bennett method for the estimation of adult mortality in developing countries. Unlike the previous method, the proposed method derives estimates of adult mortality directly from the 5-year age distribution at the mid-point of the inter-censal period. In other words, it does not convert the 5-year age distributions to single years before estimation. Results obtained from empirical studies compared favourably well with those from the original Preston-Bennett method and therefore, recommend for the estimation of adult mortality in developing countries.