This paper reports on ITER-relevant ion cyclotron resonance frequency (ICRF) physics investigated on JET in 2003 and early 2004. Minority heating of helium three in hydrogen plasmas-(He-3)H-was systematically explored by varying the 3 He concentration and the toroidal phasing of the antenna arrays. The best heating performance (a maximum electron temperature of 6.2 keV with 5 MW of ICRF power) was obtained with a preferential wave launch in the direction of the plasma current. A clear experimental demonstration was made of the sharp and reproducible transition to the mode conversion heating regime when the 3 He concentration increased above similar to 2%. In the latter regime the best heating performance (a maximum electron temperature of 8 keV with 5 MW of ICRF power) was achieved with dipole array phasing, i.e. a symmetric antenna power spectrum. Minority heating of deuterium in hydrogen plasmas-(D)H-was also investigated but was found inaccessible because this scenario is too sensitive to impurity ions with Z/A = 1/2 such as C6+, small amounts of which directly lead into the mode conversion regime. Minority heating of up to 3% of tritium in deuterium plasmas was systematically investigated during the JET trace tritium experimental campaign (TTE). This required operating JET at its highest possible magnetic field (3.9 to 4 T) and the ICRF system at its lowest frequency (23 MHz). The interest of this scenario for ICRF heating at these low concentrations and its efficiency at boosting the suprathermal neutron yield were confirmed, and the measured neutron and gammay ray spectra permit interesting comparisons with advanced ICRF code simulations. Investigations of finite Larmor radius effects on the RF-induced high-energy tails during second harmonic (omega = 2 omega(c)) heating of a hydrogen minority in D plasmas clearly demonstrated a strong decrease in the RF diffusion coefficient at proton energies similar to 1 MeV in agreement with theoretical expectations. Fast wave heating and current drive experiments in deuterium plasmas showed effective direct electron heating with dipole phasing of the antennas, but only small changes of the central plasma current density were observed with the directive phasings, in particular at low single pass damping. New investigations of the heating efficiency of ICRF antennas confirmed its strong dependence on the parallel wavenumber spectrum. Advances in topics of a more technological nature are also summarized: ELM studies using fast RF measurements, the successful experimental demonstration of a new ELM-tolerant antenna matching scheme and technical enhancements planned on the JET ICRF system for 2006, they being equally strongly driven by the preparation for ITER.