The development of tree allometric equations is crucial to accurate savannah carbon assessment. In Africa, the absence of multi-species allometric equations for savannah ecosystems has led to broad use of pan moist tropical equations develop for forest to estimate tree biomass. The aim of this study was to develop accurate multi-species allometric regressions for estimation the aboveground biomass of trees in the Western Highlands savannahs of Cameroon. Data of aboveground woody biomass were obtained from destructive sampling of 103 trees belonging to 7 species across a range of diameter (5–38 cm). The dry aboveground biomass varied from 6.19 kg for the smallest tree to 669.45 kg for the largest tree. Proportional relationships between aboveground biomass and diameter breast height (DBH) are constructed derived from eight regression models (linear, growth, compound, exponential, quadratic, cubic, power and logarithmic). The best models were selected using coefficients of determination (R²) and relative error. The best local multi-species models using only diameter as input variable were quadratic and cubic regressions with respectively R² of 0.914 and 0.917. The quadratic model overestimates the biomass with a bias of 0.04 % while the cubic model underestimates it with a 0.012 % bias. In this study, the two multi-species allometric equation developed for savannahs can be used to produce accurate estimates of biomass and carbon stocks from diameter measurement in forest inventory data.Keywords: Aboveground biomass, Allometric equations, DBH, Highlands, savannahs