Background: Mutations (mt) in the KRAS gene are common oncogenic drivers in advanced NSCLC. The KRAS mt subtypes and certain co-mutations (co-mt) have prognostic and predictive implications. New drugs promise to uniquely target KRAS G12C, raising interest in studying the unique clinical and molecular characteristics of this subtype. Methods: We retrospectively identified 56 KRAS mt advanced NSCLC patients (pts) that underwent Next-Generation Sequencing (NGS) from 06/2015 to 12/2020. We used a commercial NGS panel that tested for > 300 genomic alterations, including substitutions, insertions, deletions and copy number alterations. For statistical analysis, we divided the patients into G12C and non-G12C groups. Results: In our cohort of 56 KRAS mt pts, median age was 67 (range: 58-74) with a predominance of females (63%) and heavy smokers (89%). KRAS G12C was the most common subtype 38%; G12V 14%, G12D 11%, G12S 11%, G13D 7% and others < 5% each. G12C, G12D and G13D groups had a higher proportion of PD-L1+ tumors (84%, 100%, and 100% respectively, p = 0.02) compared to other KRAS subtypes. Pts in G12C group were on average older (median age 71 vs 61, p = 0.02) than non-G12C. Most frequent co-mt in G12C were TP53 (33%), STK11 (29%), TET2 (19%), RB1 (14%), CDKN2A/B (14%), MCL1 (14%) and ASXL1 (14%); for non-G12C, they were TP53 (54%), CDKN2A/B (37%), STK11 (34%) and RBM10 (17%). CDKN2A/B co-mt (37% vs 14%, p = 0.08) was significantly more frequent in non-G12C group and TET2 in G12C (19% vs 0%, p = 0.016). Non-G12C group more frequently had high TMB (17% vs 0%, p = 0.07) compared to G12C. No difference in survival was seen between G12C and non-G12C groups. We observed no difference in PFS (p = 0.31) or OS (p = 0.64) between smokers and no/light-smokers with KRAS mt. Co-mt with KEAP1 and SMARCA4 were significantly associated with survival in KRAS mt. Compared to KRAS+/KEAP1wt, KRAS+/ KEAP1+ pts had poor PFS (median 1.1 vs 7.5 m, p < 0.0001) and OS (1.1 vs. 27.8 m, p < 0.0001) measured from start of initial therapy. KRAS+/SMARCA4+ had worse PFS (1.0 vs 6.9 m, p < 0.0001) and OS (1.4 vs. 27.8, p = 0.0001) compared to KRAS+/SMARCA4wt. KRAS mt pts with STK11/KEAP1 that were treated with immunotherapy-based regimens had shorter PFS (1.1 vs 7.6 m, p = 0.001) and OS (1.4 vs 90.9 m, p = 0.0007) compared to those treated with chemotherapy alone. Conclusions: Our data shows that pts with KRAS co-mt with STK11/KEAP1 had worse PFS and OS with the addition of immunotherapy compared to chemotherapy alone, highlighting the potential implications of these co-mt patterns on treatment outcomes. The types of co-mts are similar between KRAS G12C and non-G12C, with the exception of CDKN2A/B (less likely) and TET2 (more likely). Larger data sets are warranted to confirm our observations and determine if these co-mts may create a predictive model for individualized therapy for KRAS mt pts, potentially independent of current predictive markers.