The most recent revision of The Denver Developmental Screening Test, the Denver II (1990), is in widespread use despite limited pre-publication validation and little published data to support its use. Review and validation studies thus far have concluded that the Denver II should not be used to determine referral for more in-depth, diagnostic evaluation, due largely to low sensitivity and specificity resulting in over-referral, excessive cost, and needless parent anxiety. However, these validation efforts have themselves been criticized for employing overly stringent criteria in diagnosing disability. The purpose of the present study was to examine the accuracy of the Denver II in identifying preterm infants with delayed cognitive development or cerebral palsy (CP). Ninety-three VLBW infants who had been transferred to a regional intensive care nursery as neonates served as subjects. Mean length of gestation and birthweight were 27.4 weeks and 1009 grams, respectively. The Denver II was administered at 4 months corrected age by a nurse certified to administer the test. At one year of age, each infant was examined by a pediatrician and was administered the Bayley Scales of Infant Development-2nd Edition by a psychologist. Both examiners were unaware of the infants' performance on the Denver II. Based on the test manual's interpretation guidelines, 67 infants were classified as Normal, 23 Questionable, and 3 Abnormal. For the 67 infants classified as normal at 4 months adjusted age (based on the Denver II), the mean and standard deviation Bayley Mental Development Index (MDI) at one year adjusted age was 96 (18), the proportion with Bayley MDI < 85 was 21%, and the proportion with CP was 12%. These same data for infants classified as either Questionable (n= 23) or Abnormal (n= 3) on the Denver II were: mean Bayley MDI 75(19), percent with Bayley MDI < 85= 15%, and percent with CP= 8%. The sensitivity of the Denver II at 4 months adjusted age for detecting infants who subsequently had a Bayley MDI < 85 was.29, while the specificity was.78. The sensitivity of the Denver II in detecting CP was.30, with specificity of.80. Results of this study suggest a lack agreement between Denver II screening and diagnostic evaluation results. Thus, while previous investigators have cautioned about over-referral rates from the Denver II, the present study suggests that under-referral may also be a problem.