BACKGROUND: While chimeric antigen receptor (CAR-T) targeting CD19 as second-line therapy for diffuse large B cell lymphoma (DLBCL) is a promising strategy, the high costs and limited access to CAR-T pose significant challenges. When assessing the cost-effectiveness of CAR-T, we need to consider not only individual outcomes but also how to effectively integrate CAR-T into the overall treatment approach for relapsed DLBCL. OBJECTIVES/STUDY DESIGN: We conducted a cost-effective analysis for patients with DLBCL in early relapse or primary refractory, to compare 'immediate CAR-T,' which proceeds directly to CAR-T, and 'late CAR-T,' which initially aims at ASCT and quickly switches to third-line CAR-T if non-responsive. The primary analysis used a patient age of 60 years, and it also examined variations from 40 to 70 years. The analysis was performed for both Japanese and US settings using a Markov model incorporating life expectancy in both countries, with extensive sensitivity analysis including factors such as age, the choice of CAR-T (lisocabtagene maraleucel or axicabtagene ciloleucel), and the opportunity to receive third-line CAR-T, to reflect real-world situations. The length of a Markov cycle was defined to be one month, and patients in the model were assumed to age one year every 12 Markov cycles. The analysis was made over a lifetime horizon, and the outcome was measured based on incremental cost-effectiveness ratio (ICER), with willingness-to-pay (WTP) thresholds of \7,500,000 and 50,000 per quality-adjusted life years (QALY) in Japan and the US, respectively, with an annual discount rate of 3%. RESULTS: Compared with 'late CAR-T,' the 'immediate CAR-T' strategy gained QALYs of 0.97 and 0.89 with an incremental cost of \5,998,354 and 8,440 in Japan and the US, respectively. The ICERs were \6,170,058/QALY in Japan and 9,596/QALY in the US. In the probabilistic sensitivity analysis for patients aged 60, 'immediate CAR-T' was cost-effective in 54.8% and 61.7% of the 10,000 Monte Carlo iterations in Japan and the US, respectively. Sensitivity analyses showed that 'immediate CAR-T' was not cost-effective when patients were over 68.4 in Japan, when the standardized mortality ratio of CAR-T and ASCT survivors was close, and when utility during treatment-free remission was low. CONCLUSIONS: Incorporating various clinical factors, the analysis showed that 'immediate CAR-T' is more cost-effective than 'late CAR-T.' However, this conclusion should be interpreted with caution, as the ICERs were very close to the WTP thresholds, and the results were highly sensitive to parameter changes.