There are two fundamentally different methods of correcting calorimetric data for heat losses to surroundings. The first method, applicable only to experiments in which the supply of heat to the calorimeter can be neglected after a certain time, uses the afterperiod cooling curve, together with a similar curve from data taken before the heat supply started, to evaluate appropriate correction terms. This problem is discussed from a general point of view and correction formulas previously reported by Regnault and Pfaundler, Schultes and Nu¨bel, Roth, and others are shown to be variations of a general scheme. A second method is needed in such experiments as the measurement of specific heat when it is not safe to neglect the heat supplied by the hot body even after a very long time. In this case, there is no afterperiod cooling curve and additional assumptions concerning the heat exchange between the calorimeter and the hot body within it must afford the needed information. The assumptions originally suggested by Rowland are re‐examined. It is shown that previous investigators (Rowland and Hoare) deduced inadequate correction methods from these assumptions. A better method is developed and a simple approximation of it is suggested. Application of the new method to a set of data gives results different from those obtained by the Rowland‐Hoare methods, from those obtained by the Regnault‐Pfaundler method (which, despite previous use, should not be applied to such data), and from those obtained by the method of Dickinson (which is recommended by the ASTM standards). The differences are significant in precision work. A brief critical review of graphical methods of obtaining the temperature correction concludes the paper. This review includes a graphical analog of the new method mentioned above.