The determination of the heat capacity in sawtooth-type, power-compensated temperature-modulated differential scanning calorimetry (TMDSC) by evaluation of the first harmonic of the Fourier transformation of the sample temperature and heatflow rate is analyzed with experiments and model calculations. At modulation periods below 100 s, the measured, uncorrected, absolute value of the complex heat capacity deviates exponentially from the true heat capacity, while the steady state within the linear heating and cooling segments is reached after about 50 s, The errors increase with the heat capacity of the sample and vary for different samples and configurations of the measurement. An empirical equation for correction is suggested. It uses a sample-mass and instrument-dependent time constant s(ln) and permits measurements down to periods of 15 s, i.e,, much before steady state is reached within the linear segments of the modulation experiment.