Telephone interviews offer an economical method of obtaining information, but little published experience addresses the use of telephone interviews for the sometimes lengthy questionnaires composed of scales with multiple-category items often required in developmental and behavioral research. In a study of the outcomes of very low birth weight infants, circumstances required that we administer a questionnaire, including seven scales composed of several Likert-type items each, to a substantial portion of the study population. Those contacted by telephone (n = 1067) differed from those responding face-to-face (n = 822) in being less likely to have a very low birth weight child and more likely to be white and of higher maternal education. The length of the interview was only slightly shorter by telephone (60.7 ± 27.9 vs 66.4 ± 21.0 minutes,p < .001), but respondent fatigue, as indicated by lower completion rates for scales at the end of the interview (92.5%) compared with those near the beginning (99.5%) did not differ by mode. Internal consistency of parental response (Cronbach's α) was high for most scales and did not differ by mode. Because assignment to mode was not random, other factors may influence our findings. However, high completion rates and comparable consistency of response supports the use of telephone interviews.J Dev Behav Pediatr 14:250–255, 1993. Index terms:health survey, child health status.