Urinary tract infection (UTI) is a very common infection. Up to every second woman will experience at least one UTI episode during her lifetime. The gold standard for identifying the infectious microorganisms is the urine culture. However, culture methods are time-consuming and need at least 24 h until the results are available. Here, we report about a culture independent identification procedure by using Raman microspectroscopy in combination with innovative chemometrics. We investigated, for the first time directly, urine samples by Raman microspectroscopy on a single-cell level. In a first step, a database of eleven important UTI bacterial species, which were grown in sterile filtered urine, was built up. A support vector machine (SVM) was used to generate a statistical model, which allows a classification of this data set with an accuracy of 92% on a species level. This model was afterward used to identify infected urine samples of ten patients directly without a preceding culture step. Thereby, we were able to determine the predominant bacterial species (seven Escherichia coli and three Enterococcus faecalis ) for all ten patient samples. These results demonstrate that Raman microspectroscopy in combination with support vector machines allow an identification of important UTI bacteria within two hours without the need of a culture step.