We examined F wave mean and minimum latency, mean and maximum amplitude, duration, persistence and chronodispersion in 241 nerves from 78 patients with Guillain-Barré syndrome (GBS) and 162 nerves from 43 patients with chronic inflammatory demyelinating polyneuropathy (CIDP). Results were compared with normal criteria derived from 72 median, 73 ulnar and 73 tibial control nerves, to determine the relative diagnostic sensitivity of the various F wave parameters. F wave abnormalities were found in 92% and 95% of nerves of patients with GBS and CIDP respectively. Absence of F responses or prolongation of minimum and mean latency were the most frequent abnormalities in both groups. Forty-five (11.2%) nerves overall had absent F responses with normal compound muscle action potential (CMAP) amplitudes and no significant fall between stimulus sites, consistent with isolated proximal conduction block. Forty-four nerves (23.7% of nerves in which F waves were present) fulfilled minimum F latency criteria for acquired demyelination . Eighty-one (20.1%) nerves had normal conventional motor nerve conduction studies and abnormal F responses, not all of which were identified by assessing only F absence or minimum latency. Severity of F wave abnormalities did not correlate with clinical outcome. Our findings confirm the high frequency of proximal nerve lesions in early GBS and CIDP, not all of which are associated with distal motor conduction abnormalities, and suggest that assessment of multiple F wave parameters, in particular chronodispersion, mean latency and mean amplitude (in addition to absence and minimum latency), increases the yield of F wave studies.