Annual surveys of leatherjacket (Tipula spp. larvae) were made in Northern Ireland to provide warning of the likelihood of damage to spring sown cereals after grass. A climate-based multiple regression model was developed to estimate mean annual populations. Ades distributions were fitted to the annual data of leatherjacket counts to provide common estimates of parameters r and r. These values were then held constant to provide yearly estimates of A, the third parameter. The relationship between A and the sample mean was established so that a frequency distribution could be generated for any estimated mean population density. The ability of these models to predict leatherjacket frequency distributions was validated by using weather data for 1985–1988 to predict the number of fields with populations in excess of thresholds of 0.5,0.75 and 1 times 105ha-1. It is concluded that the predictions were of sufficient accuracy to substitute for the annual leatherjacket survey.