In an epidemiological study, the scripts of 121 Dutch primary-school children were evaluated annually, starting in Grade 2. The children were followed for 5 years. An evaluation scale for children's handwriting (the BHK scale) was used for rating the scripts on 13 characteristics and for measuring the speed of writing. Principal component analysis of the LONG matrix (in which the data collected on the five occasions are arranged beneath each other) yielded three clusters of items: (a) fine-motor ability, (b) structural performance, and (c) stylistic preference. It was found that the children with dysgraphic handwriting (10% scoring highest on the BHK quality items) had lower fine-motor ability and, in the higher grades, showed less preference for a personal style. Their structural performance also was poorer than that of the other writers. Children with and without dysgraphic handwriting did not differ in writing speed. Some implications of the study for handwriting instruction are discussed.