A semi‐mechanistic forest growth model, 3‐PG (Physiological Principles Predicting Growth), was extended to calculate δ¹³C in tree rings. The δ¹³C estimates were based on the model's existing description of carbon assimilation and canopy conductance. The model was tested in two ∼80‐year‐old natural stands of Abies grandis (grand fir) in northern Idaho. We used as many independent measurements as possible to parameterize the model. Measured parameters included quantum yield, specific leaf area, soil water content and litterfall rate. Predictions were compared with measurements of transpiration by sap flux, stem biomass, tree diameter growth, leaf area index and δ¹³C. Sensitivity analysis showed that the model's predictions of δ¹³C were sensitive to key parameters controlling carbon assimilation and canopy conductance, which would have allowed it to fail had the model been parameterized or programmed incorrectly. Instead, the simulated δ¹³C of tree rings was no different from measurements (P > 0.05). The δ¹³C submodel provides a convenient means of constraining parameter space and avoiding model artefacts. This δ¹³C test may be applied to any forest growth model that includes realistic simulations of carbon assimilation and transpiration.