An artificial neural network model was developed to be used in a study of factors affecting the distortion of 6-to 8-mm-thick D and DH 36 grade steel plate. The data from a significant number of closely controlled welding trials, and subsequent measurements of distortion were input into the model. From this model development, a sensitivity analysis was carried out, which highlighted a number of apparently key factors, which influenced distortion. From this it was established that the carbon content of the steel plate played a key role in the amount of distortion produced by the welding process. The mechanism of the effect of carbon appears to be linked to its effect on grain size, transformation temperature, mechanical properties and pearlite content at least. It was established that an increase in carbon content was beneficial in reducing thin plate distortion caused by welding.