It has been shown that stimulation-produced-analgesia (SPA) in the cat elicited from the periaqueductal gray matter (PAG) is obtained from sites located in the ventral part, particularly the dorsal raphénucleus (DRN). These data contrast with the numerous studies performed in the rat in which efficient sites seem widely distributed throughout the PAG. These discrepancies led us to reinvestigate SPA from PAG and adjacent structures in the rat. Central stimulation was delivered through bipolar concentric electrodes (one for each animal). Analgesia was evaluated (before and during central stimulation) by measuring the modification in the vocalization threshold induced by electrical tail shocks or by considering the reaction of the animal to pinch. In contrast with the majority of previous studies, these experiments were performed on the totally freely-moving rat. The most striking result was that, in order to obtain analgesia from all regions of the PAG, it was necessary to apply intensities of central stimulation which also triggered other strong behavioral reactions. With intensities of PAG stimulation which did not induce such side effects, very few effective analgesic sites were found (21/129 sites of which 14/83 were strictly located in the PAG). However, it was possible to define two ‘pure analgesic regions’, both located in the ventral PAG: one centered on the dorsomedial part of the DRN and the other one situated in the ventrolateral PAG. No modification of nociceptive thresholds was observed when stimulating the dorsal and dorsolateral parts of the PAG as well as structures adjacent to these regions; in some rats, an increase in pain reactivity was even noted. When the intensity of central stimulation (applied to the various parts of the PAG) was increased, some stereotyped ‘behavioral responses’ occurred depending on the location of the stimulation site: motor effects (gnawing, rotation or tremor) in the ventral PAG and aversive effects (flight, jumping and on occasions, distress vocalizations) in the dorsal, dorsolateral PAG and in the ventral region just surrounding the cerebral aqueduct. Under these conditions, analgesia was obtained from practically the entire PAG, the vocalization threshold being increased dramatically on occasions. It must be emphasized that antinociceptive effects associated with other obvious behavioral manifestations (aversive ones) were also obtained from sites located outside the PAG (colliculi and tectum adjacent to the dorsal and dorsolateral PAG). These results demonstrating the major role of the ventral PAG in SPA in the rat, are in good agreement with our previous observations in the cat. In contrast, they disagree with some data reported in the literature in the rat and raise the problem of the possible involvement of ‘stress-produced-analgesia’ and motor disturbances in the appearance of most of the analgesic effects elicited by PAG stimulation.