To develop procedures for the direct dating of glacially related silts by thermoluminescence (TL), I have applied previously proposed techniques to three polymineralic 4–11 μm samples from the Toronto area. Furthermore, one of these samples was studied in detail by isolating some mineral fractions. The principal conclusion is that the TL of the lodgment Halton Till (from Woodbridge) was not significantly reset during deposition, whereas the TL of the waterlaid upper Thorncliffe silt (from the Scarborough Bluffs) was reset, as was the TL of the third sample (WC-ST), a massive silt from Woodbridge that has been generally correlated with the glaciolacustrine Sunnybrook diamict at the Bluffs. This conclusion follows from the observation of feldspar TL apparent ages of >140 ka for the ~12 ka BP Halton Till and of 36 and 66 ka for the upper Thorncliffe and WC-ST samples, respectively.The four laboratory factors that contributed most to inaccuracies in the TL apparent ages were: (1) anomalous fading in the feldspar-dominant fractions, (2) nonlinearity in TL growth curves, (3) "overbleaching" of the TL, and (4) radon loss during alpha particle counting of dry powders. The effects of each of these have been circumvented or minimized. As expected, uncertainty in the past pore-water content represented the single largest source of imprecision for such TL apparent ages after correction for the other effects.