BACKGROUND The digestive tract is the most common site of extranodal involvement in diffuse large B cell lymphoma (DLBCL) and its prognostic evaluation is different from that of ordinary DLBCL. Currently, for gastrointestinal lymphoma, in addition to the Ann Arbor staging system, the Lugano and the TNM staging systems are commonly used. However, there is no effective prognostic model to identify poor prognosis in patients with localized gastrointestinal diffuse large B cell lymphoma (GI-DLBCL). MATERIAL AND METHODS This study included 82 patients with GI-DLBCL that had a median follow-up of 75 months, and developed a model (HLAMA) with 5 variables: hemoglobin, age, lactate dehydrogenase (LDH), serum albumin, and the maximum intra-abdominal lesion diameter (MIALD). The specific indicators are: HGB <105 g/L (2 points); LDH ≥300 U/L; age ≥75 years, ALB <38 g/L, MIALD ≥4 cm (each scoring 1 point). We also developed a simplified model, which includes only 3 variables (HGB, LDH, and age). RESULTS HLAMA model and the simplified model both demonstrated good ability to predict prognosis of patients with GI-DLBCL (P<0.001), performing better than the IPI score as it could distinguish low-risk groups in relatively elderly patients (60-75 years old). CONCLUSIONS This study established a prognostic model for diffuse large B cell lymphoma of the gastrointestinal tract. Both the HLAMA model and its simplified version are similar to the IPI score, but could be considered better as they can provide a simpler and more accurate prognostic assessment in patients with GI-DLBCL. For patients with localized GI-DLBCL, our model could distinguish high-risk patients.