To protect spillways against cavitation damage, adding a small air concentration to the flow close to the invert is efficient. Aerator performance on smooth chutes was therefore well studied in terms of air entrainment and downstream air concentration development. Since bottom aerators are built upstream of regions exposed to cavitation, no damages have been observed on spillways. The introduction of roller compacted concrete (RCC) dams in the 1980s promoted the use of stepped spillways rarely used until then. Compared to conventional smooth spillways, they have the advantage of a higher energy dissipation rate, and of a self-aeration point located higher upstream. However, the non-aerated flow upstream of the inception point is exposed to an increased cavitation risk due to flow separation on the steps. Until recently, the uncertainty about the conditions required for cavitation inception motivated conservative unit discharges. Today, the cavitation potential on stepped spillways is better known and is significant, so that techniques are necessary to safely use stepped spillways under increased unit discharges. This research includes a physical model investigation of a deflector aerator on a stepped spillway. The key parameters influencing aerator performance and stepped spillway flow are systematically varied. The air concentration downstream of the aerator is measured at regularly spaced profiles by means of a fiber optical probe. The flow field downstream of the stepped spillway aerator can be described in three main zones: (i) thejet zone where air is entrained on the lower and upper surfaces, (ii) the spray and reattachment zone where spray is produced by the jet impact and where there are rapid variations of the average and bottom air concentrations, and (iii) the far-field zone where the flow depth as well as the average and bottom air concentrations gradually tend towards quasi-uniform conditions. The lower and upper surfaces of the jet issued by the deflector were considered to derive the effective takeoff angles. With the takeoff velocity, it allows to describe the lower and upper jet surfaces with projectile motion trajectories. The maximum jet elevation, the jet length, and the jet impact angle on the pseudo-bottom can then be determined. Similarly to smooth spillways, the air entrainment coefficient of the aerator is described as a function of the relative jet length. Besides, a relation for the air entrainment coefficient in function of the Froude number and the deflector geometry is presented. The average and bottom air concentration developments show a minimum shortly after the jet impact, followed by a maximum in the spray zone. These extrema are quantified and are related to the relative jet length. In the far-field zone, unlike smooth chutes, no continuous detrainment is observed for the bottom air concentration. Both the average and the bottom air concentrations gradually converge to quasi-uniform flow values. Tests with an increased approach flow bottom roughness showed a large increase of air entrainment due to the higher flow turbulence, but only small average and bottom air concentration differences downstream of the jet impact result. A pre-aerated approach flow leads to slightly higher average and bottom air concentrations downstream of the aerator. The design of a stepped spillway aerator is presented in the end to summarize the results obtained and their practical application.