Background: Diffuse large B-cell lymphoma (DLBCL) with dual expression (DE) of MYC and BCL2 has poorer prognosis than non-DE following 1^st line R-CHOP. Recent data show relapsed/refractory (r/r) DE DLBCL also has inferior outcomes after high-dose chemotherapy (CTX) and autologous stem cell transplantation (ASCT). [Herrera, et al, ASH 2015] Only 50-60% of r/r patients (pts) who receive 2^nd line CTX such as R-ICE proceed to ASCT. We studied outcomes of r/r pts treated with R-ICE and intent to transplant to determine the prognostic effect of DE. Methods: We reviewed charts of 167 consecutive de novo or transformed DLBCL pts with adequate clinical follow-up treated with 1^st line R-CHOP/R-EPOCH and salvage R-ICE from 2000 to 2015. Pts were grouped as DE (N = 26) and non-DE (N = 141). Predefined cutoffs for positivity by immunohistochemistry at relapse were ≥ 50% for MYC (N = 34) and ≥ 40% for BCL2 (N = 101). DLBCL was classified as germinal center B-cell (GC) or non-GC using Hans algorithm. Clinical characteristics were compared by Chi-square, Cochran-Mantel-Haenszel, t- or Wilcoxon tests. ASCT outcome was estimated by the cumulative incidence method and compared by the Gray test. Overall survival (OS) was estimated by the Kaplan-Meier method and compared by the log-rank test. Prognostic factors for OS were assessed by Cox proportional hazards analysis. Results: DE pts were older (mean age 63 years vs. 56 years, P = 0.009) and were less likely to be GC (44% vs. 69%, P = 0.018). 66% had ASCT, with no difference between DE (14/26) and non-DE (97/141). 5-year OS of DE pts was 26% versus 54% for non-DE (P = 0.03). There was no difference in OS between pts with or without overexpression of BCL2 or MYC in the absence of DE. In multivariable analysis, DE was associated with higher mortality risk (HR 2.23, 95% CI 1.18-4.21, P = 0.013). Conclusions: DE pts have poorer OS than non-DE pts in the 2^nd line setting when treated with R-ICE and intent for ASCT. Given the enrichment of r/r pts with BCL2 overexpression, these data provide rationale for clinical trials investigating the addition of targeted molecular therapies to R-ICE, such as the BCL2 inhibitor venetoclax, for r/r DLBCL pts who relapse after initial R-CHOP.