Based on the weekly male catches in pheromone traps and weather data for 2 years (1985, 1986), regression models were developed to predict the densities of melon fruit fly,Dacus cucurbitaeCoq., which is an important pest of cucurbit and solanaceous fruits. Among the simple linear regression models, the model SQRT (male counts + 1) = —5.37+0.75 (mean daily temperature in degrees Celsius) was the best, as it explained the highest (68%) variability. The model log(male counts + 1) = —7.54 + 0.35 (maximum temperature) + 0.63 (minimum temperature) —0.02 (maximum × minimum temperature) explained 80% variability, and was considered to be the best for forecasting fruit fly populations.