Primary tumor xenografts (PTXG) established directly from patients' primary tumors in immunosuppressed animals might represent the spectrum of histologic complexity of lung cancers better than xenografts derived from established cell lines. These models are important in the study of aberrant biological pathways in cancers and as preclinical models for testing new therapeutic agents. However, not all primary tumors engraft when implanted into immunosuppressed mice. We have investigated factors that may influence the ability of primary non-small cell lung cancer (NSCLC) to form xenografts and their association with clinical outcome.Tumor fragments from patients undergoing curative surgery were implanted into NOD-SCID (nonobese diabetic-severely combined immunodeficient) mice within 24 hours of surgery. Patient characteristics for tumors that engrafted (XG) and did not engraft (no-XG) were compared. Patient tumor DNA was profiled for the presence of 238 known mutations in 19 cancer-associated genes by using the MassARRAY platform.Xenografts were established and passaged successfully from 63 of 157 (40%) implanted NSCLCs. Tumor factors associated with engraftment included squamous histology, poor differentiation, and larger tumor size. Significantly fewer EGFR (epidermal growth factor receptor)-mutated tumors engrafted (P = 0.03); conversely, more K-RAS-mutated tumors engrafted (P = 0.05). In multivariate analysis including age, sex, stage, and mutation, patients with XG tumors had significantly shorter disease-free survival compared with no-XG patients (hazard ratio: 7.0, 95% CI: 3.1-15.81; P < 0.000003).PTXGs closely mirror the histology and molecular profiles of primary tumors and therefore may serve as important preclinical models. Tumors that engraft are biologically more aggressive and may be more representative of cancers with a higher propensity to relapse after surgery.