Thirty patients with definite or probable chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) of chronic progressive (16 patients) or relapsing (14 patients) course were randomly assigned to receive intravenous immunoglobulin (IvIg) 0.4 g per kg body weight or a placebo treatment on 5 consecutive days in a double-blind, cross-over trial. Neurological function was monitored by serial quantitative assessments [neurological disability score (NDS); clinical grade (CG) and grip strength (GS) measurements] and by electrophysiological studies before and after each treatment period. Twenty-five patients completed both treatment periods. A comparison of the observed changes in clinical outcome measures revealed statistically significant differences in favour of IvIg, with (mean +/- SD) improvements in NDS by 24.4 +/- 5.4 points (P < 0.002) in CG by 1 +/- 0.3 points (P < 0.001) in GS by +6.3 +/- 1.7 kg (P < 0.005), whereas scores were unchanged or worse with placebo. A secondary two-groups analysis of the first trial period included all 30 patients; 16 patients had been randomly assigned to IvIg and 14 to placebo treatments. Again significant differences in favour of IvIg were observed in all the clinical end-points: improvement in NDS was 35.6 +/- 25 points (P < 0.0001), in CG it was 1.3 +/- 1.9 points (P < 0.002) and in GS +9.8 +/- 7.7 kg (P < 0.001), whereas all scores worsened with placebo. Of the 30 patients, 19 (63%) improved with IvIg treatments; nine out of 16 patients (56%) with chronic progressive CIDP, and 10 out of 14 patients (71%) with relapsing CIDP (differences were not statistically significant). A placebo response was seen in five patients. Comparison of paired electrophysiological measurements before and 4 weeks after IvIg treatments revealed statistically significant improvements in the summed motor conduction velocities (sigma MCV; P < -0.0001) and in the summed compound muscle action potentials (CMAP) evoked with proximal stimulation (sigma proximal CMAP, P < 0.03) of median, ulnar, peroneal and tibial nerves. Eight of nine IvIg responders with chronic progressive CIDP improved gradually to normal function with a single 5 day course of IvIg; in five of these, small doses of prednisone were prescribed during follow-up. In 10 IvIg responders with relapsing CIDP, improvements lasted a median 6 weeks (range 3-22 weeks) and was reproducible with open label treatments. All 10 patients have been maintained and stabilized with IvIg pulse therapy of 1 g per kg body weight or less, given as a single infusion prior to the expected relapse. A beneficial response to IvIg was found to be most likely in patients with acute relapse or with disease of one year or less. Patients with predominantly sensory signs did not improve.