To define features of patients with Raynaud phenomenon that predict the evolution of connective tissue disorders, a prospective study was initiated. Patients with history or physical evidence of Raynaud phenomenon without causal or associated disorders underwent initial multisystem evaluation. Scleroderma (systemic sclerosis) of less than five years' duration was included for comparison. Patients were classified as having Raynaud phenomenon "only," undifferentiated connective tissue syndrome or scleroderma. Nailfold capillary microscopy was performed, and patterns were scored blindly from coded photographs. Of 91 patients with Raynaud phenomenon entered (Raynaud phenomenon only, n = 49; undifferentiated connective tissue syndrome, n = 22; scleroderma, n = 20), abnormal "scleroderma pattern" capillaries were noted in seven of 49, 19 of 22 and 19 of 20, respectively (p less than 0.005). Of 39 patients with Raynaud phenomenon only followed (mean duration, 23.7 months) undifferentiated connective tissue syndrome developed in three. Of 17 patients with undifferentiated connective tissue syndrome followed (mean duration, 27.6 months), six patients had transitions (four scleroderma; one scleroderma-systemic lupus erythematosus overlap; one SLE). Nailfold capillary abnormalities best identified transitional patients in both groups (eight of nine) and were more sensitive than ANA (six of nine), presence of digital ulcers (four of nine) or decreased esophageal motility (two of nine). This prospective study documents a useful role for capillary examinations in evaluating Raynaud phenomenon. Abnormal capillaries indicate an increased risk for connective tissue disease; normal capillaries favor idiopathic Raynaud phenomenon.