The sources of Galactic cosmic-ray (GCR) nuclei are still poorly identified. This is partly due to the diffusive propagation (in turbulent magnetic fields) that erases all directional information about the measured fluxes. The species that are predominantly created and accelerated in the sources are denoted 'primary species'. Those mostly created by spallation of heavier primary species are denoted 'secondary species'. While primary fluxes give access to the source parameters, secondary-to-primary ratios trace propagation processes. Key quantities for such studies are the fragmentation cross sections on the interstellar gas and the measured fluxes and ratios. In the first part of this thesis, we provide new constraints on the Galactic propagation parameters from the quartet elements (1H,2H,3He,4He), relying on a new estimate of their cross sections and an evolved statistical analysis. The derived constraints are competitive with those obtained from the standard B/C ratio analysis. The results are however limited by the precision of current measurements and motivate the development of new experiments. The second part of this thesis is dedicated to the analysis of cosmic-ray fluxes measured with the CREAM balloon-borne experiment. We present the particle identification in the detector, the estimation of efficiencies for each sub-detector, the energy reconstruction, and the atmospheric correction. Applied to the 3rd flight data (CREAM-III), we provide new data points for the boron, carbon, nitrogen and oxygen elements above 100 GeV/n.
