Truck engine waste heat recovery (WHR) systems have been investigated for many years. Among them, organic Rankine cycle (ORC) systems show the highest potential, but still lack efficient small-scale expansion devices, in practice. Proposed expanders are often of the volumetric type with oil lubrication, while partial-admission axial turbines are occasionally considered. However, despite a suitable stage pressure ratio capability and a high efficiency potential, the radial-inflow turbine remains mostly unexplored for this application, due to its high rotational speed requirement. A 8 kW ORC turbo-generator is developed based on a single-stage full-admission radial-inflow turbine and a direct-driven permanent-magnet generator. In order to handle the nominal rotational speed of 100000 revolutions per minute of the expander, the single rotating part is supported on self-acting gas bearings that are lubricated with the working fluid, i.e. halocarbon R245fa. Furthermore, in view of characterizing the turbo-generator, a fully-instrumented ORC test setup is designed and built. In a first step, the gas-bearing-supported spindle is investigated experimentally. In this configuration, the electrical machine is used in motor mode to accelerate the rotor shaft up to nominal speed, and therefore to validate the mechanical design of the rotor shaft and the bearings. Furthermore, coast-down tests are performed in order to measure the friction torque acting on the rotor shaft, both in ambient air and in the halocarbon vapor at various pressures. Experimental data and state-of-the-art model predictions are in good agreement for air. That model, however, underpredicts the measured losses by up to 36% in the halocarbon. By applying a multi-flow-regime loss model for enclosed rotating cylinder and disk, the peak deviation drops to 6.5%, suggesting non-laminar flow patterns in the gas film of the thrust bearing. In a second step, the turbo-generator is investigated experimentally on the ORC test setup. The prototype is driven up to nominal speed under off-design operating conditions, reaching a peak electrical power output of 2.3 kW and a peak overall efficiency of 67%. Despite a limited pressure ratio and the low performance of the used commercial pump, a 6% thermal efficiency is achieved on the cycle. The experimental data is compared to a reduced-order model of the turbo-generator, which is found to overpredict the prototype power output by 5 to 20%. In a last step, the validated turbo-generator model is used to assess the potential of a truck engine WHR ORC system, considering a "flat-hills" driving cycle. Under conservative assumptions, the quasi-static simulation suggests that the ORC turbo-generator could increase the powertrain power by 3.7%, and even 4.5% with a regenerative cycle. In addition to increasing the recovered power, the use of a regenerator is suggested to mitigate the cycle load fluctuations and to reduce the required size of the vapor generators and the air-cooled condenser. More experiments, with a suitable power converter, will be necessary in order to proof the ORC turbo-generator prototype up to its nominal design point, to further validate the simulation tools, and to confirm the on-board waste heat recovery potential of the technology. Nevertheless, the proposed solution already proves to be competitive compared to 1-10 kW range expanders presented in the literature.