The movement from the subjects during in vivo confocal Raman spectral measurements could change the measurement volume, leading to non-specific signals and inaccurate interpretation of the acquired spectrum. Here we introduce a generally applicable method that includes (1) developing a multimodal system to achieve real-time monitoring of every spectral measurement with reflectance confocal microscopy (RCM) and multiphoton microscopy (MPM) imaging; (2) performing region-of-interest measurement by scanning an area of the tissue during spectral acquisition. The developed method has been validated by measuring different micro-structures of in vivo human skin. Our results demonstrated great consistency between RCM images and confocal Raman spectra. The superior quality of the images and spectra allows us to derive blood flow velocity and blood glucose level. We believe this method is valuable for realizing accurate microscopic spectral measurement and have great potential to be adapted into clinic to achieve non-invasive measurement of important biological parameters.