Over 95% of Honduran coffee producers grow shaded coffee primarily with nitrogen-fixing shade trees. Some growers have begun to replace legume trees with timber trees to improve profitability. However, shade greater than 50% significantly reduces coffee yields. Knowledge of species-specific timber growth rate is needed to identify optimum densities to maximize tree growth at densities with sufficient light transmission for coffee production. The objective of this study was to identify the optimum density of Cedrela odorata (Co) and Swietenia macrophylla (Sm) in coffee agro-forestry based both on light levels to coffee and carbon accumulation in timber growth. Tree diameter and height growth were measured in 244 coffee farms with plantations ranging from 2 to 32 years in Comayagua and Santa Barbara, Honduras. In 46 fields, tree distribution and dasometic characteristics were mapped and light interception measured in a plot of 25×25 m. These data were used to parameterize a spatially explicit individual-based forest simulator model (SExI-FS) of light transmission and growth for multi-species agroforests. Simulations ranging from 3m×3m to 12m×12m showed that increasing the tree density reduced available understory light after 3 years of growth. For small farmers, wider space arrangements with 2-3 thinning appeared more suitable. With a limitation of more than 50% light transmission to coffee, the model predicts a potential volume of wood for Sm of 22-29 m3ha-1 and for Co of 28-32 m3ha-1 after 21 years. The potential carbon storage is 11 (Sm: 100 tree.ha-1) and 22 (Co: 65 tree.ha-1) Mg C ha-1.