The effects of training with overweight and underweight cricket balls on fast-bowling speed and accuracy were investigated in senior club cricket bowlers randomly assigned to either a traditional (n = 9) or modified-implement training (n = 7) group. Both groups performed bowling training three times a week for 10 weeks. The traditional training group bowled only regulation cricket balls (156 g), whereas the modified-implement training group bowled a combination of overweight (161-181 g), underweight (151-131 g) and regulation cricket balls. A radar gun measured the speed of 18 consecutive deliveries for each bowler before, during and after the training period. Video recordings of the deliveries were also analysed to determine bowling accuracy in terms of first-bounce distance from the stumps. Bowling speed, which was initially 108 +/- 5 km h(-1) (mean +/- standard deviation), increased in the modified-implement training group by 4.0 km x h(-1) and in the traditional training group by 1.3 km x h(-1) (difference, 2.7 km x h(-1); 90% confidence limits, 1.2 to 4.2 km x h(-1)). For a minimum worthwhile change of 5 km x h(-1), the chances that the true effect on bowling speed was practically beneficial/trivial/harmful were 1.0/99/< 0.1%. For bowling accuracy, the chances were 1/48/51%. This modified-implement training programme is not a useful training strategy for club cricketers.