To investigate the pathogenicity of Ureaplasma urealyticum and Mycoplasma hominis in preterm infants, we conducted a study to determine (1) frequency of isolation from cerebrospinal fluid and tracheal aspirate specimens and (2) clinical outcomes and effect of erythromycin treatment in ureaplasma-colonized infants. From the cerebrospinal fluid of 920 infants, U. urealyticum was isolated from 2 (0.2%) and M. hominis from none. From tracheal aspirate specimens from 224 infants, U. urealyticum was recovered from 37 (17%) and M. hominis from 4 (2%). Demographic characteristics and clinical outcomes were compared in very low birth weight infants (< 1500 gm) who were culture-positive or -negative for U. urealyticum. Although infants with positive results were less mature than their cohorts with negative results, there were no substantive differences in clinical outcomes between the two groups. Initiation of erythromycin treatment of infants with positive ureaplasma culture results at a mean age of 16.4 days did not appear to alter the clinical outcome. We conclude that in preterm infants (1) infection of the cerebrospinal fluid by U. urealyticum is infrequent, (2) ureaplasma organisms are frequently present in tracheal aspirate specimens but do not appear to be related to the presence or the subsequent development of respiratory disease, and (3) initiation of erythromycin treatment at 1 to 3 weeks of age does not alter the clinical course.