We use observations from a pair of 20.1 MHz riometers to constrain the magnitude and timing of the largest solar X-ray flare on record, a flare that saturated the detectors on the GOES-12 satellite. Using recordings of the sudden ionospheric disturbance associated with the flare, we show that by correcting the riometer absorption measurements for the solar zenith angle, applying a model for the effects of solar X-rays on the ionosphere, and, finally, calibrating the resulting data against the solar 0.1--0.8 nm flux from the GOES-12 satellite, we are able to closely reproduce the temporal X-ray profiles of large solar flares. We demonstrate this using two reference flares to show that we can reproduce the flare peak magnitude to approximately ¿10% and the time of the peak to better than 3 min, even when we simulate saturations of the GOES data at 50--60% of the peak flare magnitude. Analysis of the 4 November 2003 flare is complicated by the presence of radio interference near the flare peak; however, a polynomial interpolation can be used to infer the flare characteristics at the peak. By applying this analysis and considering the likely error bounds, we constrain the peak flare magnitude to lie in the range 3.4--4.8 mW/m2 (X34--X48) peaking at 1944--1948 UT. The best fit gives a peak flux of 4.0 mW/m2 (X40), peaking at 1946 UT. Taking the three independent published estimates for the peak X-ray magnitude that have associated error bounds, the weighted mean X-ray peak is found to be approximately 4.0 mW/m2 (X40), peaking close to 1946 UT.