A mechanism is proposed for the sonoluminescence (SL) arising from a single bubble maintained in levitation by an acoustic field. This proposal follows from a plasma diagnostic analysis which reveals that a single Ar bubble is characterized by a sparklike plasma (electron temperature and density: 20 000 K and 1025 m−3, respectively). The theoretical scenario (based on four hypotheses) is as follows. During its expansion and the major part of its collapse, a levitating bubble is governed by Rayleigh–Plesset dynamics. Several ns before maximum collapse the bubble interface becomes unstable and needlelike jets (the number of which is thought to be about 10) invade the bubble more or less symmetrically. Each jet (radius≈65 nm) reaches a distance equal to half of the radius of the bubble and releases a droplet (radius≈150 nm), so that an intracavity spray is released about 3 ns (perhaps less) before the time of collapse. It is implicitly proposed that the acoustic pressure at which inward jets (likely to disintegrate) may form constitutes the sonoluminescence threshold. Because of the serious distortion of the electrical double layer surrounding the bubble, both jets and droplets are electrically (but oppositely) charged. The electrical field is so high that electron emission occurs to such an extent that within 16 fs an enormous amount of energy (Joule effect) is released in the jet volume (4×1011 J m−3). The jets are ablated and a microplasma highly charged with energy (electronic temperature ⩾30 000 K; electronic density: 1026 m−3; lifetime <10 ps) expands at a velocity higher than 60 000 m s−1 so that a diverging shock wave is generated. An overpressure (higher than 1000 atm) accompanies the microplasma formation and is believed to induce strong deceleration in the bubble wall. The mechanism proposed is showed to apply to the case of bubbles in nonaqueous solvents, too. Some open questions are reported. Amongst them, (1) the possible role of the overpressure on Rayleigh–Plesset dynamics (particularly in the collapse zone) and on the aborted growth of the intracavity protuberances which might result from an incomplete ablation of the jets during the microplasma formation, (2) the effects of dissolved gases (air versus H2 or D2), (3) the effect of the liquid temperature, and (4) the role of small amounts of organic substances on SL intensity. Some experiments are suggested that may be capable of checking and distinguishing the present hypothesis from others.