OBJECTIVE: Some premature infants require prolonged tube-feeding, beyond term equivalent gestational age. Tools that could prospectively identify such infants from among otherwise healthy patients are needed. We examined how well the Neonatal Oral Motor Assessment Scale (NOMAS) predicts premature infants' transition from tube to oral feeding. DESIGN: Data from a prospective study of sucking behavior as a predictor of feeding skills were used to examine the relationship of NOMAS and other nominal predictors to pre-defined feeding milestones. SETTING: Neonatal intensive Care Unit PATIENTS: 51 tube-fed, premature infants METHODS: NOMAS was administered soon (<72 hours) after oral feeding was initiated, and weekly thereafter, until infants reached full oral feeding. A timed measure of feeding efficiency was also done as oral feeding was initiated. A standardized, permissive protocol for feeding advance was used. Nonparametric rank sum tests and ANOVA were used to relate NOMAS, feeding efficiency, and other baseline variables to feeding milestones. RESULTS: Gestational age at birth, birth weight, and initial feeding efficiency predicted shorter transition and earlier acquisition to full oral feeding; NOMAS scores did not predict feeding outcomes. CONCLUSIONS: NOMAS was a poor predictor, while feeding efficiency and other baseline traits were better predictors of feeding skills in premature infants.