The maximum urban heat island (UHI) intensity in Seoul, Korea is investigated using data measured at two meteorological observatories (an urban site and a rural site) during the period 1973-1996. The average maximum UHI is weakest in summer and strong in fall and winter. Similar to previous studies for other cities, the maximum UHI intensity is more frequently observed in the nighttime than in the daytime, decreases with increasing wind speed, and is pronounced at clear skies. A multiple linear regression analysis is performed to relate the maximum UHI to meteorological elements. Four predictors considered in this study are the maximum UHI intensity at previous day, wind speed, cloudiness, and relative humidity. It is shown that the previous-day maximum UHI intensity is positively correlated with the maximum UHI, while the wind speed, cloudiness, and relative humidity are negatively correlated with the maximum UHI intensity. Among the four predictors, the previous-day maximum UHI intensity is the most important. The relative importance among the predictors varies depending on day/night and season. A three-layer back-propagation neural network model with the four predictors as input units is constructed to predict the maximum UHI intensity in Seoul and its performance is compared with that of a multiple linear regression model. For all test data sets, the neural network model is shown to improve upon the regression model in predicting the maximum UHI intensity. The improvement of the neural network model upon the regression model is 6.3% for the unstratified test data, is higher in the daytime (6.1%) than in the nighttime (3.3%), and ranges from 0.8% in spring to 6.5% in winter.