Cortical motor neurone loss and corticospinal tract (CST) degeneration are typical of amyotrophic lateral sclerosis (ALS). It is a matter of debate whether qualitative assessment of the CST by MRI is useful in the diagnosis. It is also an open question whether quantitative determination of the T2 relaxation times can improve its value. Signal intensity along the CST on 14 consecutive slices was assessed using arbitrary visual rating on double-echo T2-weighted and proton-density spin-echo images of 21 patients with ALS and 21 age- and sex-matched controls. T2 was determined quantitatively. On the T2-weighted images the patients' ratings did not differ from that of controls. The T2 of patients and controls showed no statistical difference in any slice. There was no correlation between T2 and patient age, duration of the disease, or predominant bulbar, lower or upper motor neurone signs. The only correlation between MRI findings and disease was on the proton-density images: all cases in which the CST was poorly seen were controls; a clearly high-signal CST was seen only in the patients. High conspicuity of the CST was thus specific but not sensitive for the diagnosis of ALS. T2-weighted images and measurement of T2 were not useful for diagnosis.