Lung cancer is one of the most common malignant diseases worldwide and associated with considerable morbidity and mortality. New agents targeting the epidermal growth factor system are emerging, but only a subgroup of the patients will benefit from the therapy. Cell free DNA (cfDNA) in the blood allows for tumour specific analyses, including KRAS-mutations, and the aim of the study was to investigate the possible prognostic value of plasma mutated KRAS (pmKRAS) in patients with non-small cell lung cancer (NSCLC).Patients with newly diagnosed, advanced NSCLC eligible for chemotherapy were enrolled in a prospective biomarker trial. A pre-treatment blood sample was drawn and subsequently DNA was extracted and pmKRAS analysed. The patients received carboplatin (AUC5) i.v. day 1 and vinorelbine (30mg/m(2) i.v. day 1 and 60mg/m(2) p.o. day 8) for a maximum of six cycles. Response to chemotherapy was evaluated according to RECIST v.1.0 by CT scans of the chest and upper abdomen. The presence of pmKRAS at baseline was assessed by an in-house qPCR method. The primary endpoint was overall survival (OS). Secondary end-points were progression free survival (PFS) and overall response rate.The study included 246 patients receiving a minimum of 1 treatment cycle, and all but four were evaluable for response according to RECIST. Forty-three patients (17.5%) presented with a KRAS mutation. OS was 8.9 months and PFS by intention to treat 5.4 months. Patients with a detectable plasma-KRAS mutation had a significantly shorter OS and PFS compared to the wild type (WT) patients (median OS 4.8 months versus 9.5 months, HR 1.87, 95% CI 1.23-2.84, p=0.0002 and median PFS 3.0 months versus 5.6 months, HR 1.60, 95% CI 1.09-2.37, p=0.0043). A multivariate Cox regression analysis confirmed the independent prognostic value of pmKRAS in OS but not in PFS. The response rate to chemotherapy was significantly lower in the group of patients with a mutation compared to WT (p<0.0001).The presence of KRAS mutations in plasma may be a marker of poor prognosis and may also hold predictive value. Further validation in an independent cohort is highly needed.