In the applications of thermoluminescence (TL) in dosimetry and archaeological dating, it is usually assumed that the measured TL depends on the total dose applied and it is independent of the dose rate. Thus, calibration of a TL specimen is usually performed at a significantly higher dose rate than that of the dose to be determined. A few experimental accounts in the literature report on dose-rate dependences of TL intensity for a given total dose. One theoretical work published gave a numerical solution of the simultaneous differential equations governing the filling of traps and centres at different dose rates. In the present work, the numerical solution is extended so that it includes the other important stage of TL, namely the heating phase. It is shown that with a rather simple model of one trapping state and two kinds of recombination centres, dose-rate effects may occur. An appropriate choice of the relevant parameters indeed yields one emission, which increases with increasing dose rate, whereas another emission decreases with the dose rate with a constant total dose, in agreement with an experimental result in quartz quoted in the literature.