The purpose of this investigation was to identify the findings of the studies devoted to the child voice, most of which have occurred in the past 25 years, and to present a synthesis of these findings with respect to the pedagogy, or art and science, of teaching children to sing. The data suggest that a philosophical disparity exists about singing, that is, teachers believe that it is an easy and natural process, whereas voice scientists support the contrary point of view. Data about children indicate that they have large ranges as infants, begin to sing with greater accuracy in pitch contour than to specific pitches, and gradually increase their range and pitch-matching accuracy as they mature. Many studies reported differences in accuracy with respect to sex, although the vocal mechanisms are virtually identical until puberty. While a variety of results exist for using text with pitch matching, for using accompaniment, for specific ranges at age levels, and for using certain vocal models, there are implications for teaching. Based on the majority of findings, it is suggested that teachers use particular activities with initial singers to help them hear their voices and learn to control the mechanism; that children learn to match one note, then short patterns, before progressing to songs; that the ranges of the children should determine the selected literature to create an effective match; and that vocal modeling be carefully done to assist in helping children match pitch and timbre. The most important element is that children can be taught to sing, suggesting that they have the mechanism in place and the teacher must properly develop their skills in singing.