A retrospective metabolomics approach was utilized as a tool in genomics-assisted selection for forest tree improvement. Key metabolic components were thus correlated with the inherently rapid stem growth of Pinus densiflora Siebold & Zucc. trees, including water soluble metabolites and amino acids found in the inner bark (phloem) of cambial region tissues that were harvested in midsummer (July) from 34-year-old trees. Metabolites were assessed in individual genotypes from seven open-pollinated, half-sibling families. Four of the families were classed as fast growing, and three of the families were significantly slower growing. This metabolomics approach was also used to assess metabolite profiles in similar cambium–phloem region tissues from 4-year-old trees representing 12 unrelated families. Initially by Pearson’s correlation analysis, and subsequently by stepwise linear modeling, we assessed the interactions between stem growth parameters (stem diameters and a stem volume index) and metabolite contents, and we did this via a retrospective approach. Among the metabolites identified, inositol was a consistent, positive, and highly significant correlate with the stem growth of P. densiflora trees at two developmental stages (ages 4 and 34 years). Inositol also exhibited these significant correlations in an environment-independent manner, i.e., on two very different field progeny test sites. We conclude that inositol may be a useful metabolic biomarker for early selection of rapid tree stem growth traits of this conifer species. In so doing, it could appreciably enhance the breeding and selection of inherently rapid-growing families of P. densiflora.