Multimodel inference of length‐at‐age and tag–recapture data was used to estimate growth for western population segment (WPS; populations natal to the Pearl River, Louisiana, and the Pascagoula River, Mississippi) Gulf Sturgeon Acipenser oxyrinchus desotoi. Four candidate growth models were fitted to the length‐at‐age data (n = 174): a three‐parameter von Bertalanffy growth function (VBGF), a two‐parameter VBGF, a power model, and a Gompertz model. Model support for each data set was estimated using Akaike's information criterion. The Gompertz model provided the best fit to our length‐at‐age data and had the lowest mean asymptotic maximum length (L∞). Mean estimated growth parameters for WPS Gulf Sturgeon were similar to those estimated for eastern population segment Gulf Sturgeon, but some of the mean parameter estimates fell outside of our 95% confidence intervals. Three growth curves were also fitted to tag–recapture data (n = 116): a reformulated two‐parameter VBGF, the Francis () GROTAG model, and a reparametrized Gompertz model. The GROTAG model provided the best fit to our data; all three models had lower mean estimates for L∞ compared with parameter estimates based on length at age but had similar mean growth coefficients. We also compared the observed growth of 47 individuals with their predicted growth based on parameter estimates of the Gompertz length‐at‐age model and their age at tagging. We found that observed growth was generally less than what was predicted for juveniles and adults. For adults, we attribute this to the inaccuracies associated with age estimation for older individuals, which suggests that a tag–recapture approach might be the most appropriate method of modeling Gulf Sturgeon growth. Our study (1) demonstrates the importance of fitting multiple models to estimate growth, (2) will enable more accurate growth comparisons for Gulf Sturgeon across their range, and (3) represents the most robust length‐at‐age and tag–recapture data set for WPS Gulf Sturgeon.