Monitoring processes underlying mitochondrial homeostasis is crucial for understanding of the cell energetics. Nevertheless, we still lack appropriate non-invasive methods capable to evaluate these processes directly in living systems. In the last decades, time-resolved spectroscopy of endogenous fluorescence has been shown a perspective approach for non-invasive investigation of metabolic changes. Combination of time-resolved fluorescence measurement and imaging, known as Fluorescence Lifetime Imaging (FLIM), provides a deeper insight into natural mitochondrial metabolism and has thus been proven as an advanced diagnostic and imaging tool. We studied cell autofluorescence, derived from endogenous fluorophores present in living cells naturally, without need of fluorescent dyes or probes. To evaluate fluorescence lifetimes in combination with fluorescence spectra and/or images, we have applied Time-correlated Single Photon Counting (TCSPC), after excitation with pulsed picosecond lasers. We focused at fluorescence derived from NAD(P)H (excitation 375nm) and flavins (excitation 450-477nm) to characterize changes in the mitochondrial metabolic state in living cells. Activators and inhibitors of the respiratory chain were applied to quantify recorded metabolic changes. Different cell types were compared, ranging from cardiac cells to cancer cell lines. Separation of individual components from complex fluorescence signals is crucial for correct analysis of measured data. Spectral characteristics, unique for each fluorophore, are one way of fluorophore identification in complex biological samples. However, spectra of endogenous fluorophores are often overlapping, which is complicating data analysis. Time-resolved fluorescence decay patterns are additional effective means of separation of distinct fluorophores. We demonstrate application of advanced analytical approaches for the the extraction of individual fluorescence components from the recorded time-resolved spectroscopy datasets, including the spectral linear unmixing, principal component analysis and blind source separation. Presented study of the time-resolved spectroscopy and imaging of mitochondrial metabolic state in living cells has great potential to improve several biomedical applications, including early diagnostics and treatment of complex diseases, which are often linked with modifications in cellular energetics.