Fatigue cracks grow from pores at the surface of components that were produced by laser powder bed fusion (LPBF). In AlSi10Mg components produced by LPBF, large oxides apparently interfere with consolidation of powder into the melt pool, contributing to part porosity; the oxides may also nucleate hydrogen porosity. In previous work, it was found that the effect of such porosity on fatigue life could be predicted by measuring pores found on a sample size of a few square millimeters and extrapolating to the much larger surface of a fatigue test specimen. The aim of this work is to understand the fundamental origin of oxides in LPBF as a basis for controlling the defects. The sources considered here are the native oxide on the surface of metal powder and oxidation of hot spatter in the build chamber for the case of LPBF of UNS N07718 samples. Kinetic analysis indicates that the rate of oxidation of a spatter droplet would be controlled by the oxygen concentration in the build chamber. From measurement of the surface coverage of deposited oxide particles (apparently oxidized spatter) on the build surface, and estimating the thickness of these deposits, it is concluded that about twice as much oxidized spatter is deposited on the part surface (during building of each layer) than the amount of oxygen incorporated into the part from this source. A possible reason for this difference is that spatter oxides might be partially removed from the part surface during recoating.