The aim of this work is to assess the performance of multimodal spectroscopic approach combined with single core optical fiber for detection of bladder cancer during surgery in vivo. Multimodal approach combines diffuse reflectance spectroscopy (DRS), fluorescence spectroscopy in the visible (405 nm excitation) and near-infrared (NIR) (690 nm excitation) ranges, and high-wavenumber Raman spectroscopy. All four spectroscopic methods were combined in a single setup. For 21 patients with suspected bladder cancer or during control cystoscopy optical spectra of bladder cancer, healthy bladder wall tissue and/or scars were measured. Classification of cancerous and healthy bladder tissue was performed using machine learning methods. Statistically significant differences in relative total haemoglobin content, oxygenation, scattering, and visible fluorescence intensity were found between tumor and normal tissues. The combination of DRS and visible fluorescence spectroscopy allowed detecting cancerous tissue with sensitivity and specificity of 78% and 91%, respectively. The addition of features extracted from NIR fluorescence and Raman spectra did not improve the quality of classification. This study demonstrates that multimodal spectroscopic approach allows increasing sensitivity and specificity of bladder cancer detection in vivo. The developed approach does not require special probes and can be used with single-core optical fibers applied for laser surgery.