The experiment studied the effects of a short duration exposure to traumatic memories using magneto-encephalography (MEG). Nine right-handed DSM-4 PTSD patients were recruited from a unit for anxiety disorders and an organisation supporting victims of violence. In order to have a homogeneous sample, we included only women who suffered from civilian PTSD. Exclusion criteria were co-morbid major medical illness, metallic dental prostheses that would interfere in the magnetic measurement, and current drug treatment. All participants were free from neurological disease and had normal hearing. They signed a written informed consent form. An ethics committee accepted the study. A tape-recorded voice administered a script-driven imagery. The patients had to imagine, successively, a neutral image, a traumatic memory and rest, while MEG measured brain activities across delta, theta, alpha and beta bands. Each condition lasted three minutes. Heart rate (HR), anxiety and the vividness of mental images were recorded at the end of each phase. MEG power analysis was carried out with Statistical Parametric Mapping (SPM) 8. The signals were averaged for each of the three conditions of threeminutes duration. The dependent variable was a subtracted value: (trauma - rest) - (neutral - rest). The significance threshold was set at P<0.01. Anxiety and HR significantly increased during the trauma condition and returned to the neutral level during rest. The vividness of the mental imagery remained stable across the three conditions. The left-brain demonstrated a statistically significant power decrease in the secondary visual cortex (BA 18-19) in the delta band, the insula (BA13) in the beta band, the insula (BA13), premotor cortex (BA 6), Broca area (BA 44), and BA 43, in the alpha band. The symptom provocation protocol was successful in eliciting subjective anxiety and HR response in relation to traumatic memories. Our MEG results are in keeping with previous neuro-imagery studies showing decreased activities in the insula and Broca area during PTSD symptom provocation. However, we did not replicate the activation in the amygdala and the cingulate and prefrontal cortex found in some studies. Moreover, the within-group design, the small sample, and the inclusion of only female patients with milder dissociative symptoms limit our conclusions. The MEG protocol we used may also explain some partial discrepancies with previous MEG studies. However, our aim was to provoke a specific autobiographic recall of a traumatic event unfolding several sequential mental images along three minutes as in exposure therapy for PTSD. Despite its limitations, this pilot study is the first to provide MEG data during trauma recall. It suggests that recalling a specific traumatic event along three minutes results in hypo-activations of the brain regions regulating language and emotions. This paves the way to recording whole sessions of specific therapies for PTSD, with MEG using the millisecond resolution. MEG might be of interest to study the suppression of traumatic memories and their activation and habituation through prolonged graduated exposure in imagination across several sessions. MEG could also be used to study the effects of medication on PTSD symptoms. A controlled replication in a larger sample including male and female patients with various traumatic experiences is needed.