Forty children were studied as they began to learn to read at 5 years old and as they developed their skill up to 7 years old. At each year they were tested for ability on 44 variables which measured ability in reading, spelling, vocabulary, short-term memory, visual perception and discrimination, auditory-visual integration, language knowledge, phonological awareness, grammatical knowledge, rote knowledge and ordering ability, and performance on the Weschler Intelligence Scale for Children. The data for the whole group of children were analysed in several ways: (1) At each year the patterns of associates of reading skill were determined (with and without control for IQ); (2) the abilities at one point in time which were associated with later reading skill were charted for the whole group, for a subset of children who at 5 years old started with no reading skill, and for another group of children who were progressing rapidly at 7 years old. Cross-lagged correlation comparisons were made to investigate causal paths. These analyses allowed us to chart the course of reading development and the interactive ways in which associated skills such as spelling, reading, phonological awareness and syntactic knowledge grow from each other differentially at different stages of development. The nature of reading skill changes rapidly in the first 3 years of acquisition. In information processing terms it begins as an undifferentiated skill associated with knowledge of the letters of the alphabet, phonological awareness, and visual symbolic short-term memory processes. It then changes in character, being associated with holistic visual pattern recognition skills. By 6 years old phonological awareness and verbal short-term memory processes are by far the strongest associates. By 7 years old the better readers' skills are associated with analytic visual perceptual analysis, the learning of new symbol-sound associations, and sound blending skill. Reading has become a multifaceted ability tapping a wide range of different skills from language comprehension to analysis of the order of elements in a visual array.Our present knowledge of reading is assembled in reviews of thousands of individual studies of reading (e.g. Gibson and Levin, 1975;Mitchell, 1982;Vellutino, 1979;Vernon, 1971). The preponderance of these studies have been ex post facto bivalent designs with little or no attempt to look for differential abilities. They have been performed by different investigators, with children of different cultures, education, age, socioeconomic background and intelligence, and they have involved radically different numbers of subjects. They have taken place over the past 50 years when educational practices have been changing. It is quite possible that the conclusions of the reviews therefore constitute a nomothetic generality which, from a heterogeneous population, reflects none of the individuals studied. Such review efforts may also fail to lead to an understanding of the development of reading. If we want to study development we must d...