The load forecast using optimized support vector machine was proposed based on accumulated temperature effect.The correction model of temperature was established based on study of two forms of accumulated temperature effects. In order to improve prediction accuracy, optimized support vector machine was adopted to forecast load. Finally, with the data of a region in Jiangsu as history data, least squares was adopted to obtain optimal parameters, and then the correction of temperature was used into the above method to forecast load. The results show high prediction accuracy and have good prospect.