The present study investigated the possibility of clinical application of somatosensory cerebral evoked response (SER) in the diagnosis of the site and level of various neural lesions (peripheral and central). SERs were recorded from the hand-sensory area of the scalp by electrical stimulation of the median nerve at the wrist and an averaging procedure (digital computer). The study group comprised 43 healthy subjects and 267 patients clinically having the following neural lesions: peripheral nerve lesion (50 patients); cervical nerve root lesion (17); spinal cord lesion (52); brain stem/cerebellum lesion(33); internal capsule/thalamus/basal ganglia lesion (53); frontal cortex lesion (14); frontopariental cortex lesion(33); or parieto-temporal cortex lesion(25). Abnormal SER wave patterns were classified into 11 types by comparison with both normal SER patterns obtained from the healthy subjects and the clinical diagnoses. The results were as follows: 1. Normal SER waves always consisted of a pattern of 5 peaks (3 negative and 2 positive: N1, P1, N2, P2, N3). 2. Abnormal SER wave patterns were classified into 11 types (author's classification): "Type A", prolonged latency of N1; "Type B", low amplitude of all peaks; "Type C", flat pattern without peaks; "Type Da", lack of N1; "Type Db", N1 present but other waves absent; "Type Dc", N2 absent; "Type Dd", high amplitude of N2; "Type De", high amplitude of N3; "Type Df", high amplitude of N1; "Type Ea", unilateral high amplitude of all peaks; and "Type Eb", high amplitude of all peaks bilaterally. 3. In peripheral nerve, cervical nerve root or spinal cord lesions, SERs were usually Type A, B, or C and reflected the degree of sensory impairment. SERs showed little variation in patients having motor nerve impairment only. In internal capsule/thalamus/basal ganglia lesions, SERs were usually Type B or C. In contrast to the results for peripheral nerve lesions, this was true not only in patients with sensory impairments but also in those who had motor impairment only. Type De was detected in some patients with motor impairment only. Such variations in the SER pattern seemed to be derived from the site and grade of lesion. In cortex lesions, SER often showed Type Dc, De, or Ea unrelated to whether the impairment was sensory or motor in nature. Type Dc was characteristic of lesions associated with the motor cortex, implying that this area was the origin of the N2 peak. 4. Abnormal SERs were detected in 58.8% of all the 267 patients. 46.9% of patients with motor impairment only and 77.5% of patients with more than moderate sensory impairment had abnormal SERs. 5. The origins of the components of the SER pattern were thought to be as follows: N1 and P1, direct primary response to the stimulus via the specific sensory projection system; N2 and P2, association areas near the sensory-motor cortex; the late component N3, non-specific sensory projection system. These results for SERs suggest that clinical application of the SER may prove extremely useful in diagnosing the site and level of neural lesions.