Claudin 18.2 (CLDN18.2) is a tight-junction protein overexpressed and differently exposed in solid tumors such as gastric cancer (GCa), gastroesophageal junction adenocarcinoma, and pancreatic ductal adenocarcinoma (PDAC). We sought to explore the potential of CLDN18.2 as a biomarker for molecular imaging and targeted radiopharmaceutical therapy in GCa and PDAC models. Methods: Bulk and single-cell RNA sequencing for CLDN18.2 from 31 patients with PDAC were performed. Subcutaneous xenografts of GCa with the GSU cell line and PDAC with HUPT-4 and PATU8988S cell lines were developed in nude mice. Serial PET imaging with 89Zr-labeled zolbetuximab, an anti-CLDN18.2 monoclonal antibody ([89Zr]Zr-DFO-zolbetuximab), and human IgG ([89Zr]Zr-DFO-IgG) as control (5.5-7.4 MBq) was performed 1, 3, and 6 d after injection, followed by ex vivo analysis of biodistribution. Tumor-bearing mice received a single intravenous injection of [177Lu]Lu-DOTA-zolbetuximab (7.4 or 14.8 MBq), nonradiolabeled zolbetuximab, [177Lu]Lu-DOTA-IgG, [177Lu]Lu-DOTA, or saline as the control. Toxicity of [177Lu]Lu-DOTA-zolbetuximab was evaluated by laboratory and histologic analyses over 90 d. Results: RNA sequencing confirmed significantly higher expression of CLDN18.2 in human PDAC compared with noncancerous pancreas tissue, with no significant difference between treated and untreated tumors. Serial PET imaging demonstrated a high tumor uptake of [89Zr]Zr-DFO-zolbetuximab at all 3 time points (mean uptake on day 6: 24.4 ± 7.8 %ID/g in GSU; 36.6 ± 10.1 %ID/g in PATU8988S; 16.48 ± 4.7 %ID/g in HUPT-4), and significantly higher than the uptake seen in mice imaged with [89Zr]Zr-DFO-IgG (day 6: 4.8 ± 1.9 %ID/g in GSU; P = 0.0029). The highest tumor-to-background uptake ratio was achieved on day 6 (GSU tumor-to-muscle ratio, 14.85 ± 7.8). Biodistribution analyses were consistent with the PET imaging results. High-dose [177Lu]Lu-DOTA-zolbetuximab (14.8 MBq) resulted in reduced tumor growth in GSU and PATU8988S over 4 and 8 wk, respectively, with complete regression of most HUPT-4 tumors and improved 90-d survival compared with the mice treated with control conditions. The 90-day treatment toxicity assay indicated a favorable safety profile. Conclusion: CLDN18.2 could serve as a promising biomarker for precise quantitative imaging and effective treatment of GCa and PDAC. This proof-of-concept study encourages the clinical translation of CLDN18.2 as a radiotheranostic in patients with CLDN18.2-expressing tumors.