This study consists of a detailed analysis of the analgesic effects induced by stimulation of the various parts of the periaqueductal gray matter (PAG) in the freely moving rat. In order to characterize the analgesia, two criteria are considered: (1) the evaluation of the degree of analgesia and behavioral side effects evoked during central stimulation; and (2) the presence of post-effects. Central stimulation (50 Hz sine waves) was delivered via bipolar concentric electrodes and analgesia was quantified by the change in the vocalization threshold induced by electrical stimulation of the tail. Within the ventral PAG, the vocalization threshold increased gradually with the intensity of the central stimulation, the degree of analgesia generally being powerful. There was no relationship between the strength of the analgesic effects and the motor disturbances also produced by stimulation of this region. Antinociceptive effects generally disappeared when the stimulation ceased. Only when the intensity of the stimulation was strong enough to induce very powerful analgesic effects were post-stimulation analgesic effects noticed. Within the dorsal and dorsolateral PAG as well as in the ventral region just surrounding the aqueduct, analgesia appeared suddenly, was generally less pronounced and was always concomitant with strong aversive reactions. In contrast with the analgesia from the ventral PAG, marked post-effects were observed. These latter characteristics were also obtained from stimulation of regions located outside the PAG (colliculi, intercollicular commissure and tectum adjacent to the dorsolateral PAG) although these zones were not extensively studied. By consideration of various data in the literature, it is concluded from this study, which clearly distinguishes stimulation-produced-analgesia (SPA) from ventral PAG versus dorsal PAG, that analgesia induced from this midbrain area involves at least two different neuronal substrates. Whilst the ventral PAG seems to be more preferentially involved in pain modulation, the authenticity of 'analgesia' triggered by stimulation of aversive regions (which are widely spread over the PAG) is questioned and proposals to explain the simultaneous appearance of analgesic effects and aversion are considered.