Source segregated domestic food waste (FW) collected by Veolia Environmental Services was used as a test substrate for therrnophilic anaerobic digestion in a laboratory-scale trial with semicontinuous feeding at an organic loading rate of 3 g of volatile solids (VS) L-1 day(-1), reached after an acclimatization period of 60 days. The FW had a total Kjeldahl nitrogen content of 0.73% on a wet weight basis which, without dilution, gave a digestate total ammonia nitrogen (TAN) concentration of >5.0 g of N L-1. This led to the accumulation of propionic and other longer-chain volatile fatty acids (VFA), followed by a sharp increase in acetic acid, which was sufficient to overcome the ammonia buffering capacity, causing the pH to fall and gas production to cease. In digesters run in parallel, TAN concentrations were regulated by diluting the FW with water. This allowed the determination of critical threshold concentrations for TAN, and for free ammonia nitrogen (FAN) by calculation, and monitoring of the pattern of VFA production. Below 2.5 g of N L-1 of TAN, there was no evidence of digestion instability. Between 2.5 and 3.5 g of N L-1 of TAN, transient peaks in VFA could be seen, but without long-term accumulation; above 3.5 g of N L-1 of TAN, continuous accumulation of VFA occurred, which eventually led to failure. Stable digestion could be maintained by a dilution of 0.5:1 water to FW. The energy implications of using a dilution strategy were evaluated using the ADAT modeling tool for a range of scenarios at dilution ratios of 0.5-3:1, ambient temperatures of 5-35 degrees C, and different plant sizes. As expected, dilution had the largest effect on the net energy demand, but at the lowest dilution required to overcome toxicity, this only equated to a 2-6% loss of the raw energy potential of the biogas produced from the FW when compared to mesophilic digestion with pasteurization.