Abstract Canada’s boreal zone covers almost 60% of Canada’s landmass. It has large areas of peatlands where black spruce is the most common tree species, which rarely reaches merchantable size. Tree biomass is usually estimated using allometric equations with diameter at breast height (DBH) used as a predictor. While such allometric relationships are well-established for merchantable trees, predicting biomass for the sub-merchantable trees poses a challenge due to scarcity of destructive samples and suitable predictors, as many trees may be shorter than 1.3 m. Mensuration and destructively sampled biomass component data were collected from 679 trees at 58 black spruce peatland sites in the Boreal Plains and Taiga Plains ecozones. These data were used to develop individual tree allometric models, stand level total volume yield curves, and total volume-tobiomass conversion models for Boreal Plains, Taiga Plains and the two ecozones combined. Allometric biomass models performed well for stemwood (88–97% variance explained) and bark (89–95%), but less well for branches (65–82%) and foliage (42– 70%). Basal diameter–based models outperformed diameter at breast height, especially when combined with height, and the selected yield model reliably reproduced observed trends and extrapolated realistically beyond observed conditions. Implementation in CaMP increased aboveground small-tree biomass carbon (C) stocks by about 40% and deadwood—by about 27%, however increase in tree biomass and deadwood had little effect on peat C stocks, highlighting the importance of mosses and other understory vegetation on peat accumulation. Updated parametrization significantly increased NEP across sites (26– 87 g C m−2 yr−1; p<0.05) with improved agreement with eddy flux data.