Many authors consider that late onset is a suspect criterion for differentiating primary Raynaud's phenomenon (Raynaud's disease, RD) from Raynaud's syndrome (RS). However, many cases of late-onset Raynaud's phenomenon in patients over forty years of age remain without etiologic diagnosis and therefore deserve the designation "late-onset RD." One hundred and ninety-four patients with RD (143 women, 51 men) were selected among 424 patients with Raynaud's phenomenon, according to Allen and Brown's criteria with negative serologic investigations and normal capillaroscopy. The purpose of the study was to consider the possible discriminant value of age of onset in distinguishing between RD and RS. The following epidemiologic features were studied: age of onset, sex, family history of Raynaud's phenomenon and migraine, and smoking and working habits. Microcirculation was assessed by capillaroscopy and strain-gauge plethysmography. Maximal digital flow at 45 degrees C and reactivity to cold were determined for each patient. Results were related to age of onset. The existence of true cases of late-onset RD in patients over forty years of age was confirmed (prevalence 27%), showing a correlation with a family history of Raynaud's phenomenon inferior to that found in early-onset cases (p < 0.0001). Microcirculation studies generally indicated a strong correlation between reactivity to cold, familial RD, and early onset, whereas no correlation was found with migraine or smoking. Nor was there any clinical or plethysmographic evidence for arteritis as a possible pathogenetic factor in late-onset RD. These results indicate that late-onset RD is a valid designation and that its pathogenesis seems less dependent on genetic sensitivity to cold than that of early-onset cases. In the absence of underlying arteritis, neurovascular dysfunction or a hemorheologic mechanism may be suggested as plausible causes of late-onset RD.