Parallel factor analysis (PARAFAC) is a decomposition method that can decompose multilinear data into unique components consisting of loading vectors and score vectors. Here, it was used to decompose trilinear excitation-emission matrices of fluorescing dissolved organic matter (FDOM) into individual components. Maxima of excitation and emission loadings of individual components can be determined and used to identify the broad chemical composition of FDOM and its origin based on published literature. This study investigated the spectral properties of colored dissolved organic matter (CDOM) and FDOM from the central Arctic Ocean. I assessed which fluorescent components could be identified using PARAFAC and I examined the vertical distribution of CDOM and FDOM characteristics in the central Arctic Ocean. Absorption coefficient at 375 nm (a375) had an average value of (mean±SE) 0.37±2.08×10-2 m-1. The ratio between the absorption coefficients at 250 nm and 365 nm, E2:E3, was on average 13.7±0.84, and the spectral slope coefficient between 275 and 295 nm, S275-295, was22.5±0.65 µm-1. a375 and E2:E3 were highest in surface waters and decreased with depth in the form of a power function, whereas S275-295 increased with depth. Based on the relationship ofa375 vs. the spectral slope coefficient between 300 and 650 nm, S300-650, CDOM was classified as being of terrestrial origin. The absorption coefficient at 300 nm was used to differentiate between freshly-produced semi-labile CDOM and refractory CDOM. Semi-labile CDOM was detected down to a water depth of 1,500 m. A PARAFAC model for three components could be validated and explained 99.0% of the data variability. Fluorescent component FC1 had an excitation loading maximum at <240 nm and an emission loading maximum between 412 and 416 nm. Based on these maxima, the component could be identified as a terrestrial, humic-like component that is characteristic for forest streams and wetlands. The second fluorescent component FC2 had an excitation loading maximum at 255 nm and a maximum emission loading at 454 nm. It was again identified as a terrestrial, humic-like component from forest streams and wetlands and is a degradation product of humic substances. The third fluorescent component that was identified with PARAFAC had an excitation loading maximum at 280 m and an emission loading maximum at 340 nm. It has similar spectral properties like the classical peak T and could be tryptophan-/ protein like. When the information about CDOM and FDOM were combined in a principal component analysis, loadings of the three absorption coefficients at 300 nm, 350 nm, and 375 nm were positively correlated with principal component 1, whereas loadings of the spectral slope coefficient 2S275-295 and the maximum fluorescence intensity of FDOM component FC2 were positively correlated with principal component 2. As the scores of surface water masses clustered around the loadings of the absorption coefficients, principal component 1 might represent the terrestrial derived, freshly-produced CDOM in surface waters. The fluorescence intensity of FC 1 decreased with depth, whereas the fluorescence intensity ofFC2 was relatively stable, but highest in deep waters. This suggests that terrestrial-humic likeFC1 is more and more degraded with increasing water depth and its degradation product FC2accumulates in deep waters. The fluorescence intensity of tryptophan-like/ protein-like FC3 had a maximum fluorescence intensity in intermediate waters (~150–2,000 m) and this component was likely released during a phytoplankton bloom. This study shows that CDOM and FDOM properties have a vertical distribution in the central Arctic Ocean and can be used to distinguish between surface water and the water masses below.