4091 Background: Liver resection of patients with colorectal cancer liver metastasis (mets) may provide curative therapy. The decision for resection is based on CT- or MRI scans and includes technical aspects (esp. remaining liver tissue) and prognostic factors (i.e. number of metastases). Methods: During two workshops, CT- or MRI scans at baseline (BL) and after 4 months treatment (follow-up, F-U) of CELIM-patients (pts) were presented to surgeons (surg.) who were blinded to each other, the time of investigation (BL or F-U) and clinical data. Scans were allocated to one of the following categories: (1) resection, (2) exploration, (3) borderline resectable (prior chemotherapy preferred) or (4) non-resectable If scans were allocated to category (3) or (4), the reason (technical or poor prognosis) was stated. For evaluation of resectability over time, pts were regarded as resectable according to presented imaging, if ≥ 50% of surg. voted for (1) or (2). Inter-observer variability was evaluated by a reduced model, too: operation (1+2), borderline, primary chemo therapy preferred (3) and non-resectable, chemotherapy (4). Critical disagreements were assumed if ≥1 surg. voted for (4) and ≥ 50% surg. for (1+2) or vice versa. Results: Seven surg. (≥ 5/workshop) participated at the review. A total of 180 scans were reviewed. Both, BL and F-U scans, were available for 75 pts (68% of study pts). During the review, 24 pts changed from ‘non-resectable‘ at BL to ‘resectable‘ at F-U, 5 pts from ‘resectable‘ to ‘non-resectable‘, 29 pts remained ‘non-resectable‘, and 17 pts ‘resectable‘ (19 pts more resectable at F-U; chi-square: p=.021). R0 resection was actually performed in 16/29 pts classified as ‘resectable‘ (55%) and 13/54 pts not classified as ‘resectable‘ (24%) according to presented BL- images. In the full model (6 possibilities), the agreement between surg. was 51.4 %, p<.001. In the clinically more important reduced model, the agreement was 63.5% (p<.001). Critical disagreements were observed in 18 cases (10%). Conclusions: Resectability according to scans increased significantly during the CELIM trial. Regarding the clinical approach, different surgeons had a high degree of agreement. [Table: see text]