Objective: To investigate the efficacy and safety differences between the cisplatin + paclitaxel (TP) and cisplatin + fluorouracil (PF) regimens in combination with or without immune checkpoint inhibitors (ICIs) in advanced esophageal squamous cell carcinoma (ESCC) first-line treatment and prognostic factors. Methods: We selected the medical records of patients with late stage ESCC admitted to the hospital between 2019 and 2021. Based on the first-line treatment regimen, control groups were divided into chemotherapy plus ICIs ( n = 243) and non-ICIs ( n = 171), 119 (49%) in the TP + ICIs group, 124 (51%) in the PF + ICIs group, 83 (48.5%) in the TP group, and 88 (51.5%) in the PF group in the control group. We analyzed and compared factors related to efficacy, safety, or response to toxicity and prognosis across four subgroups. Results: The overall objective response rate (ORR) and disease control rate (DCR) of the TP plus ICIs group were 42.1% (50/119) and 97.5% (116/119), respectively, which were 6.6% and 7.2% higher than those of the PF plus ICIs group. Patients in the TP combined with ICIs group had higher overall survival (OS) and progression-free survival (PFS) than those in the PF combined with ICIs group [hazard ratio (HR) = 1.702, 95% confidence interval (CI): 0.767–1.499, p = 0.0167 and HR = 1.158, 95% CI: 0.828–1.619, p = 0.0055] ORR and DCR were 15.7% (13/83) and 85.5% (71/83) in the TP chemotherapy alone group, significantly higher than the PF group [13.6% (12/88) and 72.2% (64/88)] ( p < 0.05), OS and PFS were also better in patients treated with TP regimen chemotherapy than PF (HR = 1.173, 95% CI: 0.748–1.839, p = 0.0014 and HR = 0.1.245, 95% CI: 0.711–2.183, p = 0.0061). Furthermore, following the combination of TP and PF diets with ICIs, the OS of the patients was higher than that of the group treated with chemotherapy alone (HR = 0.526, 95% CI: 0.348–0.796, p = 0.0023 and HR = 0.781, 95% CI: 0.0.491–1.244, p < 0.001). Regression analysis showed that the neutrophil-to-lymphocyte ratio (NLR), the control nuclear status score (CONUT), and the systematic immune inflammation index (SII) were independent prognostic factors for the efficacy of immunotherapy ( p < 0.05). The overall incidence of treatment-associated adverse events (TRAEs) was 79.4% (193/243) and 60.8% (104/171) in the experimental and control groups, respectively, and there was no statistically significant difference in TRAEs between TP + ICIs (80.6%) and PF + ICIs (78.2%) (61.4%) and PF groups (60.2%) ( p > 0.05). Overall, 21.0% (51/243) of patients in the experimental group experienced immune-related adverse events (irAEs), and all of these adverse effects were tolerated or remitted following drug treatment without affecting follow-up. Conclusion: The TP regimen was associated with better PFS and OS with or without ICIs. Furthermore, high CONUT scores, high NLR ratios, and high SII were found to be associated with poor prognosis in combination immunotherapy.