A new approach is presented for cell lysate identification which uses SERS-active silver nanoparticles and a droplet-based microfluidic chip. Droplets are generated by injecting silver nanoparticles, KCl as aggregation agent, cell lysate containing cell constituents, such as nucleic acids, carbohydrates and proteins into a continuous flow of mineral oil. This platform enables accurate mixing of small volumes inside the meandering channels of the quartz chip, and allows acquiring thousands SERS spectra with 785 nm excitation at an integration time of 1 second. Preparation of three batches of three leukemia cell lines demonstrated the experimental reproducibility. A classification model based on support vector machine was built based on SERS spectra. Leave-one-batch out cross validation gave sensitivities, specificities and accuracies better than 99% to differentiate Jurkat, THP-1 and MONO-MAC-6 leukemia cell lysates. This approach is compared with previous published reports and an outlook is given for transfer to intact cells.