We used portfolio theory to analyze the tradeoffs between returns and performance instability of deployment units for Pinus radiata D. Don. We considered three groups of 34 trees each grown to produce appearance lumber, structural lumber, or both. Risk was based on the variability of tree returns in scenarios of changing volume, wood stiffness and presence of resin defects due to genotype by environment interaction inducing both changes of scale and differential tree response to environmental scenarios. The return of structural trees was highly variable with a mean of 3.11 NZ $/stem/year, followed by appearance-structural trees (3.48 NZ $/stem/year). In contrast, appearance trees had the lowest returns (1.99 NZ $/stem/year) and variability. The portfolio model selected structural trees in high-risk scenarios, but selection was apportioned between structural and appearance-structural trees as the risk decreased. The model selected only appearance trees for high-risk aversion. The analysis also considered silvicultural regimes, where the appearance-structural regime was selected under high variability. As risk decreased the appearance grades regime was also selected. The structural regime was rarely selected due to the variability of stiffness between trees. Using genotypes improved for stiffness could increase the expected value and reduce variability for structural purposes, making the structural regime more appealing.