In latent growth curve modeling (LGCM), overall fit indices have garnered increased disputation for model selection, and model fit evaluation based on the mean structure has becoming popularity. The present study developed a versatile fit index, named Weighted Root Mean Squared Errors (WRMSE), based on individual case residuals (ICRs) with the aim of facilitating model selection. Through comparing to various fit indices, this study demonstrated that WRMSE could provide effective model fit information beyond previous indices. Due to higher sensitivity to misspecification in the mean structure, WRMSE was suggested to be utilized as a supplement for the traditional overall fit indices in the process of model selection. The limitation of WRMSE was that for complex models beyond the commonly used polynomial models, an increased number of repeated measurement occasions was required to ensure the precision of WRMSE.