We used a stratified random sampling design to inventory the mangrove vegetation within the Zambezi River Delta, Mozambique, to provide a basis for estimating biomass pools. We used canopy height, derived from remote sensing data, to stratify the inventory area, and then applied a spatial decision support system to objectively allocate sample plots among five strata. Height and diameter were measured on overstory trees, saplings and standing dead trees in nested plots, and biomass was calculated using allometric equations. Each of the eight mangrove species occurring in Mozambique exist within the Delta. They are distributed in heterogeneous mixtures within each of the five canopy height classes, not reflecting obvious zonation. Overstory trees averaged approximately 2000 trees ha⁻¹, and average basal area ranged from 14 to 41 m² ha⁻¹ among height classes. The composition of the saplings tended to mirror the overstory, and the diameter frequency distributions suggest all-aged stands. Above-ground biomass ranged from 111 to 483 Mg ha⁻¹ with 95 % confidence interval generally within 15 % of the height class mean. Despite over 3000 trees ha⁻¹ in the small-tree component, 92 % of the vegetation biomass is in the overstory live trees. The objective inventory design proved effective in estimating forest biomass within the 30,267 ha mangrove forest.