Paraneoplastic cerebellar degeneration (PCD) is a heterogeneous group of disorders characterized by subacute cerebellar ataxia, specific tumour types and (often) associated antineuronal antibodies. Nine specific antineuronal antibodies are associated with PCD. We examined the relative frequency of the antineuronal antibodies associated with PCD and compared the neurological symptoms and signs, associated tumours, disability and survival between groups of PCD with different antibodies. Also, we attempted to identify patient-, tumour- and treatment-related characteristics associated with functional outcome and survival. In a 12-year period, we examined >5000 samples for the presence of antineuronal antibodies. A total of 137 patients were identified with a paraneoplastic neurological syndrome and high titre (> or =400) antineuronal antibodies. Fifty (36%) of these patients had antibody-associated PCD, including 19 anti-Yo, 16 anti-Hu, seven anti-Tr, six anti-Ri and two anti-mGluR1. Because of the low number, the anti-mGluR1 patients were excluded from the statistical analysis. While 100% of patients with anti-Yo, anti-Tr and anti-mGluR1 antibodies suffered PCD, 86% of anti-Ri and only 18% of anti-Hu patients had PCD. All patients presented with subacute cerebellar ataxia progressive over weeks to months and stabilized within 6 months. The majority of patients in all antibody groups had both truncal and appendicular ataxia. The frequency of nystagmus and dysarthria was lower in anti-Ri patients (33 and 0%). Later in the course of the disease, involvement of non-cerebellar structures occurred most frequently in anti-Hu patients (94%). In 42 patients (84%), a tumour was detected. The most commonly associated tumours were gynaecological and breast cancer (anti-Yo and anti-Ri), lung cancer (anti-Hu) and Hodgkin's lymphoma (anti-Tr and anti-mGluR1). In one anti-Hu patient, a suspect lung lesion on CT scan disappeared while the PCD evolved. Seven patients improved by at least 1 point on the Rankin scale, while 16 remained stable and 27 deteriorated. All seven patients that improved received antitumour treatment for their underlying cancer, resulting in complete remission. The functional outcome was best in the anti-Ri patients, with three out of six improving neurologically and five were able to walk at the time of last follow-up or death. Only four out of 19 anti-Yo and four out of 16 anti-Hu patients remained ambulatory. Also, survival from time of diagnosis was significantly worse in the anti-Yo (median 13 months) and anti-Hu (median 7 months) patients compared with anti-Tr (median >113 months) and anti-Ri (median >69 months). Patients receiving antitumour treatment (with or without immunosuppressive therapy) lived significantly longer [hazard ratio (HR) 0.3; 95% confidence interval (CI) 0.1-0.6; P = 0.004]. Patients > or =60 years old lived somewhat shorter from time of diagnosis, although statistically not significant (HR 2.9; CI 1.0-8.5; P = 0.06).