We studied 50 patients with definite, probable, and possible multiple sclerosis (MS), prospectively (20 patients) and retrospectively (30 patients), to determine the value of the hot bath test for diagnosing MS and to compare it to visual evoked responses and oligoclonal bands. The hot bath test was abnormal in 8 of the 23 patients with definite MS (35%), and 4 of the 27 patients with probable or possible MS (15%). Only one patient with an abnormal hot bath test did not also have other evidence of definite multiple sclerosis. Our results suggest that the hot bath test seldom adds diagnostic information, especially when tests for evoked responses and oligoclonal bands are available.