Data from the 10-year follow-up of the 1970 British Births Survey were examined for associations between motor performance and dyslexia. Five tests of motor performance were used: (a) balancing on one leg, (b) throwing a ball in the air, clapping and catching it, (c) walking backwards, (d) sorting matches and (e) graphaesthesia (recognizing shapes drawn on the palm of the hand). These tests were given to 12 950 children aged between 10 and 11 years old. The cohort was divided into nine groups based on three levels of literacy achievement and three levels of possible indicators of dyslexia. The group with the most severe underachievement and most possible indicators (children most likely to be severely dyslexic) comprised about 2% of the total. Of this group, 35.3% failed one motor test and 16.4% failed more than one (51.7% in total), compared with 26.8% and 7.7% of normal achievers. The children had greater problems with balance than those in the other two severely underachieving groups but the effects were small. It is suggested that the use of a balance test only as a screener for dyslexia could result in a proportion of dyslexics being missed and that remedial motor training programmes for children with dyslexia should be offered only to those with co-occurring motor difficulties.