A basic problem intrinsic to many clinical diagnostic procedures as well as minimally invasive surgeries is the online invivo classification of tissue. Associated with this problem is the task to determine the boundaries between tissue sections of various degrees of disease progression, which cannot be identified easily. This problem is partly founded in the imaging modalities conventionally used, i.e., white-light endoscopy or fluorescence-based endoscopic imaging. These techniques allow for extracting of only a limited parameter set for judging the physiological or pathological state of tissue. Furthermore, fluorescence-based endoscopy relies on the administration of external labels, which principally disturbs the native tissue. These problems can be circumvented using Raman microspectroscopy as a diagnostic tool. Raman microscopy allows to record vibrational spectra at each sampling point. Therefore the molecular fingerprint of the sample can be deciphered with spatial resolution. It has been shown that Raman spectroscopy in combination with advanced statistical methods can be used to identify and grade tissue samples. However, the conventional approach of judging excised tissue sections by Raman microscopy does not present an approach which can be readily used in the clinics. Here we present our recent progress towards designing a fiber-based Raman probe, which - in perspective - might be incorporated into the working channel of a surgical endoscope. Thereby, it is anticipated to contribute to the clinical routine. We will review the general design principle of such a device and the specific design strategy for our Raman probe in concert with comparative measurements employing a set of home-built and commercially-available devices.© (2011) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.