Temporal processing deficit could be associated with a specific difficulty in learning to read. In 1951, Stambak provided preliminary evidence that children with dyslexia performed less well than good readers in reproduction of 21 rhythmic patterns. Stambak's task was administered to 1,028 French children aged 5-6 years. The score distribution (from 0 to 21) was quasi-normal, with some children failing completely and other performing perfectly. In second grade, reading was assessed in 695 of these children. Kindergarten variables explained 26% of the variance of the reading score at second grade. The Stambak score was strongly and linearly related to reading performance in second grade, after partialling out performance on other tasks (oral repetition, attention, and visuo-spatial tasks) and socio-cultural level. Findings are discussed in relation to perceptual, cerebellar, intermodal, and attention-related theories of developmental dyslexia. It is concluded that simple rhythm reproduction tasks in kindergarten are predictive of later reading performance.