Circulating tumour cells (CTCs) are cells which have disseminated from the primary tumour site and invaded the peripheral blood. Since CTCs are derived from the primary tumour, they have potential use as prognostic markers. They could also provide insight into tumour phenotype and predict response to various treatment modalities without the need for biopsy. As CTCs are vastly outnumbered by normal blood cells (tens of tumour cells amongst millions of white blood cells), the development of CTC isolation and enrichment techniques is paramount to the successful use of CTCs in research applications. In this small pilot study, we tested the recovery and isolation efficiency of the ClearCell FX microfluidic label-free separation system (ClearBridge Biomedics) against our current separation technique - RosetteSep CTC enrichment antibody cocktail with CD36 (Stemcell Technologies) to determine whether this new approach would be suitable for isolating CTCs from the blood of patients with non-small cell lung cancer (NSCLC). Normal donor blood was spiked using one of five fluorescently-labelled (CellTracker Green) NSCLC cell lines, as a model for CTCs, and processed in parallel using each separation platform. The recovered cells were fixed onto a microscope slide, and the whole field containing the fluorescently-labelled tumour cells and non-labelled contaminating blood cells was imaged using an Olympus VS120 slide digitiser. Recovery of spiked tumour cells scored from the images showed the yields were comparable for both techniques (average recovery 70% and 57% across the cell lines for ClearCell FX and RosetteSep, respectively) with considerable variation between the cell lines. Both methods reduced the number of blood cells by around 4 orders of magnitude (>99.9% reduction), with the ClearCell FX system showing lower numbers of non-target cells on average. The enrichment and purity observed using the ClearCell FX system suggests that this platform is suitable for the label-free recovery of CTCs from the blood of patients with NSCLC, and is comparable to our existing antibody-based negative selection method.