Gastrointestinal (GI) cancers are among the most fatal diseases in the world. Numerous studies have demonstrated the relationship between autophagy and development of gastrointestinal cancers. However, whether autophagy-related genes can predict prognosis of GI cancers in individuals of Asian ancestry has not been defined. This study, evaluated the prognostic value of autophagy-related genes in gastrointestinal cancer. Expression profile of autophagy-related genes for 296 gastrointestinal cancer patients of Asian ancestry was downloaded from the TCGA database (TCGA-LIHC, TCGA-STAD, TCGA-ESCA, TCGA-PAAD, TCGA-COAD, TCGA-CHOL, and TCGA-READ). The prognostic value of the autophagy-related genes was evaluated using univariate Cox, LASSO, and multivariate Cox regression analyses. The risk score of the autophagy-related gene signature was calculated to assess its predictive prognostic value for GI cancers. Forty-seven differentially expressed autophagy-related genes, in Asian patients with gastrointestinal cancers, were identified. Of the 47 genes, 4 were associated with prognosis of GI cancer (SQSTM1, BIRC5, NRG3, and CXCR4). A prognostic model for GI cancer, based on the expression of the above 4 genes in the training set, showed that cancer patients were stratified into high-risk and low-risk groups (
P
<
0.05
). The utility of the model for overall survival (OS) of GI cancer patients was consistent across the entire set, training set, and test set (entire set:
P
=
4.568
Ă
10
â
4
; train set:
P
=
5.718
Ă
10
â
3
; test set:
P
=
3.516
Ă
10
â
2
). The sensitivity and specificity of the ROC curve of the above prognostic model in predicting the 5-year prognosis of GI cancer was satisfactory (entire set: 0.728; train set: 0.727; test set: 0.733). Analysis of clinical samples validated the overexpression of the 4 genes (SQSTM1, BIRC5, NRG3, and CXCR4) in tumor tissues relative to paired normal tissues, consistent with bioinformatic findings. Expression of the 4 autophagy-related genes (SQSTM1, BIRC5, NRG3, and CXCR4) can accurately predict the prognosis of gastrointestinal tumors in Asian patients.