The shea tree, Vitellaria paradoxa, shields people, crops and livestock in West African parkland agroforestry systems from climate variability. Accurate estimates of accumulated biomass of such key species may support ways to secure financial incentives within global climate policies. In this quest, variation in allometric relations used for biomass (carbon stock) estimates on the basis of stem diameter matters, but parameters a and b of the standard format (AGB = a Db; AGB = aboveground biomass, D = stem diameter) are correlated and are directly related to tree shapes. Functional branch analysis (FBA) allows non-destructive derivation of allometrics. For Koutiala and Yanfolila shea populations empirical branching parameters were scale-independent, matching FBA assumptions. Allometry (AGB = 169 (D/20)2.64 and AGB = 146 (D/20)2.65, kg tree−1) implied 22%, 16% and 11% larger vegetative aboveground biomass in Koutiala than in Yanfolila at stem diameters (D) of 10, 20 and 32 cm, respectively. Below-ground biomass predictions (BGBi = 8.73 (Di/10)2.35 and BGBi = 8.16 (Di/10)2.38, kg per proximal root) differed- 6% and- 15% for root diameters Di of 10 and 32 cm, respectively. On a dry weight basis, the shoot:root ratio was 2.7 and 2.9, respectively for the two sites. Stand-level above-and below-ground carbon stocks in Koutiala (2.16 ± 0.44 and 0.8 ± 0.15 Mg C ha−1) were not significantly different from those in Yanfolila (3.21 ± 0.60 and 1.26 ± 0.21 Mg C ha−1), respectively. Further research is required to include (potential) fruit production to the plant architectural model.