Knowledge regarding neural pain processing is primarily the result of studies involving models of brief cutaneous pain; however, clinical pain generally originates in deep tissue and is prolonged. This study measured the dynamic neural activation associated with a muscular pain model incorporating both acute and tonic states. Hypertonic saline (5% NaCl) was infused into the brachioradialis muscle of eleven healthy volunteers for 15min after an initial bolus of 0.5mL. Ten controls followed the same protocol with normal saline (0.9% NaCl). Magnetic resonance images of cerebral blood flow (CBF) were acquired using an arterial spin labelling method. The imaging volume extended from the thalamus to the primary somatosensory cortices, but did not include the brainstem and cerebellum. Using a numerical scale from 0 to 10, ratings of pain intensity peaked at 5.9+/-0.6 and remained near 5 for the remainder of the trial. Controls experienced minimal pain, reporting a peak value of 1.8+/-0.4. Significant CBF increases in rostral and caudal anterior insula bilaterally, anterior mid-cingulate cortex (aMCC), bilateral thalamus, and contralateral posterior insula were observed. The time courses of CBF revealed significant differences in the activation pattern during tonic pain. In particular, a more rapid return to baseline in aMCC versus insula was interpreted as a preferential decrease in the affective component of pain. This conclusion was supported by the strong correlation between pain intensity ratings and CBF in the contralateral insula (R(2)=0.911, p<0.01), which is a region believed to be responsible for pain intensity processing.