The goal of this study was to estimate the accuracy of thoracic paraspinal (T-PSP) electromyography (EMG), performed at the time of initial evaluation, for the subsequent clinical diagnosis of amyotrophic lateral sclerosis (ALS). Medical records from a consecutive series of patients referred for suspected ALS were abstracted. Those in whom T-PSP EMG was not performed or for whom follow-up was not available were excluded. The study population included 64 patients. T-PSP EMG was abnormal in 28 (44%) study patients, 26 (93%) of whom were subsequently diagnosed with ALS. T-PSP EMG showed neither fibrillation potentials nor positive sharp waves in 36 (56%) patients, 9 (25%) of whom later developed ALS. The sensitivity of T-PSP for the diagnosis of ALS was 0.74 and the specificity was 0.93. The likelihood ratio positive was 10.5 and the likelihood ratio negative was 0.28. These results suggest that T-PSP EMG is very useful for "ruling in" the diagnosis of ALS, but less useful as a screening test for "ruling out" the diagnosis. Because T-PSP EMG is used clinically in an effort to confirm the diagnosis (and not as a screening test), these results suggest that T-PSP EMG is an extremely useful adjunct for the diagnosis of ALS.