The cosmic-ray hydrogen and helium spectra have been measured by the Balloon Borne Experiment with a Superconducting Solenoid Spectrometer (BESS), which has been flown from Lynn Lake, Manitoba, Canada, annually since 1993. The BESS experiment provides excellent rigidity measurement and precise particle identification with a large geometric acceptance. We present here the hydrogen and helium nuclei energy spectra from 0.2 to 10 GeV nucleon-1 and their isotopic composition from 0.2 to about 1 GeV nucleon-1 for the first BESS flight. This provides the first simultaneous measurements of the cosmic-ray secondaries, deuterons, and 3He, with their primaries, protons, and 4He over this energy range in a period of solar minimum. In this paper, we have achieved significant improvements in data analysis in the following aspects. First, the latest available cross-section data and their parameterizations were utilized in the simulation code developed for this study. Second, a complete simulation was performed for both protons and heavy ions: the δ-ray effect was properly simulated and showed a large influence on the measurement of heavy ions at high energies. Third, the secondary particle correction, which dominates the systematic uncertainty at low energies for singly charged particles, protons and deuterons, was calculated iteratively with the simultaneously measured primary cosmic-ray spectra. In general, the results of this experiment are consistent with other recent measurements using balloon-borne or satellite experiments, but with better precision. The measured spectra of protons, deuterons, 3He, and 4He and their corresponding ratios are compared with different interstellar/heliospheric propagation calculations, which were derived to fit observations of heavy nuclei. The overall good agreement indicates that the propagation history for light cosmic-ray elements, protons, deuterons, and helium nuclei is similar to that of the heavy nuclei. The 2H/1H ratio is sensitive to the propagation models, and our results show a tendency of better agreement with the reacceleration model than the standard leaky-box model.