We assessed (1)H-MRS as a screening tool for detection of hippocampal sclerosis in patients with temporal lobe epilepsy (TLE). (1)H-MRS was carried out in the hippocampus of 23 patients with unilateral TLE. Metabolite alterations detected by (1)H-MRS correlated with degree of segmental neuronal cell loss and amount of astrogliosis. Positive correlation was found between total N-Acetylaspartate (tNAA) reduction and neuronal density in hippocampal CA1 (P < 0.001), CA3 (P = 0.015), and CA4 subfields (P = 0.031) and the dentate gyrus (P = 0.006). Neuronal cell loss in CA1 turned out to be the most predictive and only significant variable for tNAA reduction (P = 0.027). The association between myo-inositol (m-Ins) and astroglial glial fibrillary acidic protein (GFAP) expression revealed significantly increased m-Ins concentrations associated with diffuse astrogliosis (m-Ins = 6.4 +/- 1.1 institutional units) compared with gliosis restricted to isolated sectors of the hippocampus (i.e. hilus) (m-Ins = 5.2 +/- 1.2 institutional units) (P = 0.039). A negative correlation was found between m-Ins and neuronal loss in the CA4 subfield of the hippocampus (P = 0.028). Our results support (1)H-MRS as a suitable non-invasive method for preoperative identification of hippocampal sclerosis in patients with TLE. The extent of tNAA reduction correlates with hippocampal neuronal cell density. Furthermore, m-Ins is associated with the extent of hippocampal astrogliosis.